| { |
| "paper_id": "O11-2008", |
| "header": { |
| "generated_with": "S2ORC 1.0.0", |
| "date_generated": "2023-01-19T08:05:37.217527Z" |
| }, |
| "title": "Feature Analysis of Chinese Textual Entailment System", |
| "authors": [ |
| { |
| "first": "Wan-Chi", |
| "middle": [], |
| "last": "\u9ec3\u6587\u5947", |
| "suffix": "", |
| "affiliation": {}, |
| "email": "" |
| }, |
| { |
| "first": "Shih-Hung", |
| "middle": [], |
| "last": "Huang", |
| "suffix": "", |
| "affiliation": {}, |
| "email": "" |
| }, |
| { |
| "first": "", |
| "middle": [], |
| "last": "Wu", |
| "suffix": "", |
| "affiliation": {}, |
| "email": "" |
| }, |
| { |
| "first": "Liang-Pu", |
| "middle": [], |
| "last": "\u9673\u826f\u5703", |
| "suffix": "", |
| "affiliation": {}, |
| "email": "" |
| }, |
| { |
| "first": "\u8c37\u5733", |
| "middle": [], |
| "last": "Chen", |
| "suffix": "", |
| "affiliation": {}, |
| "email": "" |
| }, |
| { |
| "first": "Ku", |
| "middle": [], |
| "last": "Tsun", |
| "suffix": "", |
| "affiliation": {}, |
| "email": "" |
| }, |
| { |
| "first": "", |
| "middle": [], |
| "last": "\u8cc7\u8a0a\u5de5\u696d\u7b56\u9032\u6703", |
| "suffix": "", |
| "affiliation": {}, |
| "email": "" |
| } |
| ], |
| "year": "", |
| "venue": null, |
| "identifiers": {}, |
| "abstract": "", |
| "pdf_parse": { |
| "paper_id": "O11-2008", |
| "_pdf_hash": "", |
| "abstract": [], |
| "body_text": [], |
| "back_matter": [], |
| "bib_entries": { |
| "BIBREF0": { |
| "ref_id": "b0", |
| "title": "Probabilistic textual entailment: Generic applied modeling of language variability", |
| "authors": [ |
| { |
| "first": "Ido", |
| "middle": [], |
| "last": "Dagan", |
| "suffix": "" |
| }, |
| { |
| "first": "Oren", |
| "middle": [], |
| "last": "Glickman", |
| "suffix": "" |
| } |
| ], |
| "year": 2004, |
| "venue": "Proceedings of the Workshop on Learning Methods for Text Understanding and Mining", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Ido Dagan and Oren Glickman, Probabilistic textual entailment: Generic applied modeling of language variability, In Proceedings of the Workshop on Learning Methods for Text Understanding and Mining, Grenoble, France, 2004.", |
| "links": null |
| }, |
| "BIBREF1": { |
| "ref_id": "b1", |
| "title": "NTCIR 9", |
| "authors": [], |
| "year": null, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "NTCIR 9, Recognizing Inference in TExt task, http://artigas.lti.cs.cmu.edu/rite/Main_Page.", |
| "links": null |
| }, |
| "BIBREF2": { |
| "ref_id": "b2", |
| "title": "A survey of paraphrasing and textual entailment methods", |
| "authors": [ |
| { |
| "first": "Ion", |
| "middle": [], |
| "last": "Androutsopoulos", |
| "suffix": "" |
| }, |
| { |
| "first": "Prodromos", |
| "middle": [], |
| "last": "Malakasiotis", |
| "suffix": "" |
| } |
| ], |
| "year": 2010, |
| "venue": "Journal of Artificial Intelligence Research", |
| "volume": "38", |
| "issue": "", |
| "pages": "135--187", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Ion Androutsopoulos and Prodromos Malakasiotis, \"A survey of paraphrasing and textual entailment methods\", Journal of Artificial Intelligence Research, Volume 38, pages 135-187, 2010.", |
| "links": null |
| }, |
| "BIBREF3": { |
| "ref_id": "b3", |
| "title": "New ranking algorithms for parsing and tagging: Kernels over discrete structures, and the voted perceptron", |
| "authors": [ |
| { |
| "first": "Michael", |
| "middle": [], |
| "last": "Collins", |
| "suffix": "" |
| }, |
| { |
| "first": "Nigel", |
| "middle": [], |
| "last": "Duffy", |
| "suffix": "" |
| } |
| ], |
| "year": 2002, |
| "venue": "Proceedings of the 40th Annual Meeting on Association for Computational Linguistics", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Michael Collins and Nigel Duffy, \" New ranking algorithms for parsing and tagging: Kernels over discrete structures, and the voted perceptron\", Proceedings of the 40th Annual Meeting on Association for Computational Linguistics, 2002.", |
| "links": null |
| }, |
| "BIBREF4": { |
| "ref_id": "b4", |
| "title": "Recognising textual entailment with logical inference", |
| "authors": [ |
| { |
| "first": "Johan", |
| "middle": [], |
| "last": "Bos", |
| "suffix": "" |
| }, |
| { |
| "first": "Katja", |
| "middle": [], |
| "last": "Markert", |
| "suffix": "" |
| } |
| ], |
| "year": 2005, |
| "venue": "Proceedings of the conference on Human Language Technology and Empirical Methods in Natural Language Processing", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Johan Bos, Katja Markert, \"Recognising textual entailment with logical inference\", Proceedings of the conference on Human Language Technology and Empirical Methods in Natural Language Processing, Vancouver, B.C., Canada, 2005.", |
| "links": null |
| }, |
| "BIBREF5": { |
| "ref_id": "b5", |
| "title": "COGEX at RTE 3", |
| "authors": [ |
| { |
| "first": "Marta", |
| "middle": [], |
| "last": "Tatu", |
| "suffix": "" |
| }, |
| { |
| "first": "Dan", |
| "middle": [], |
| "last": "Moldovan", |
| "suffix": "" |
| } |
| ], |
| "year": 2007, |
| "venue": "Proceedings of ACL-PASCAL Workshop on Textual Entailment and Paraphrasing", |
| "volume": "", |
| "issue": "", |
| "pages": "22--27", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Marta Tatu and Dan Moldovan, \"COGEX at RTE 3\", In Proceedings of ACL-PASCAL Workshop on Textual Entailment and Paraphrasing, pages 22-27, Prague, Czech Republic, 2007.", |
| "links": null |
| }, |
| "BIBREF6": { |
| "ref_id": "b6", |
| "title": "A semantic approach to recognizing textual entailment", |
| "authors": [ |
| { |
| "first": "Marta", |
| "middle": [], |
| "last": "Tatu", |
| "suffix": "" |
| }, |
| { |
| "first": "Dan", |
| "middle": [], |
| "last": "Moldovan", |
| "suffix": "" |
| } |
| ], |
| "year": 2005, |
| "venue": "Proceedings of HLT/EMNLP 2005", |
| "volume": "", |
| "issue": "", |
| "pages": "371--378", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Marta Tatu, Dan Moldovan, \"A semantic approach to recognizing textual entailment\", In Proceedings of HLT/EMNLP 2005, pages 371-378, Vancouver, Canada, 2005", |
| "links": null |
| }, |
| "BIBREF7": { |
| "ref_id": "b7", |
| "title": "WordNet: An Electronic Lexical Database", |
| "authors": [ |
| { |
| "first": "Christiane", |
| "middle": [], |
| "last": "Fellbaum", |
| "suffix": "" |
| } |
| ], |
| "year": 1998, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Christiane Fellbaum, \"WordNet: An Electronic Lexical Database\", The MIT Press, 1998.", |
| "links": null |
| }, |
| "BIBREF8": { |
| "ref_id": "b8", |
| "title": "Logic form transformation of WordNet and its applicability to question answering", |
| "authors": [ |
| { |
| "first": "Dan", |
| "middle": [ |
| "I" |
| ], |
| "last": "Moldovan", |
| "suffix": "" |
| }, |
| { |
| "first": "Vasile", |
| "middle": [], |
| "last": "Rus", |
| "suffix": "" |
| } |
| ], |
| "year": 2001, |
| "venue": "Proceedings of the 39th Annual Meeting of ACL", |
| "volume": "", |
| "issue": "", |
| "pages": "402--409", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Dan I. Moldovan and Vasile Rus, \"Logic form transformation of WordNet and its applicability to question answering\", In Proceedings of the 39th Annual Meeting of ACL, pages 402-409, Toulouse, France, 2001.", |
| "links": null |
| }, |
| "BIBREF9": { |
| "ref_id": "b9", |
| "title": "An information-theoretic definition of similarity", |
| "authors": [ |
| { |
| "first": "Dekang", |
| "middle": [], |
| "last": "Lin", |
| "suffix": "" |
| } |
| ], |
| "year": null, |
| "venue": "Proceedings of the Fifteenth International Conference on Machine Learning", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Dekang Lin, \"An information-theoretic definition of similarity\", In Proceedings of the Fifteenth International Conference on Machine Learning (ICML-98)", |
| "links": null |
| }, |
| "BIBREF11": { |
| "ref_id": "b11", |
| "title": "Dependency-based construction of semantic space models", |
| "authors": [ |
| { |
| "first": "Sebastian", |
| "middle": [], |
| "last": "Pad\u00f3", |
| "suffix": "" |
| }, |
| { |
| "first": "Mirella", |
| "middle": [], |
| "last": "Lapata", |
| "suffix": "" |
| } |
| ], |
| "year": 2007, |
| "venue": "Computational Linguistics", |
| "volume": "33", |
| "issue": "2", |
| "pages": "161--199", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Sebastian Pad\u00f3 and Mirella Lapata, \"Dependency-based construction of semantic space models\", Computational Linguistics, Volume 33, No. 2, pages 161-199, 2007.", |
| "links": null |
| }, |
| "BIBREF12": { |
| "ref_id": "b12", |
| "title": "Vector-based models of semantic composition", |
| "authors": [ |
| { |
| "first": "Jeff", |
| "middle": [], |
| "last": "Mitchell", |
| "suffix": "" |
| }, |
| { |
| "first": "Mirella", |
| "middle": [], |
| "last": "Lapata", |
| "suffix": "" |
| } |
| ], |
| "year": 2008, |
| "venue": "Proceedings of the 46th Annual Meeting of ACL: HLT", |
| "volume": "", |
| "issue": "", |
| "pages": "236--244", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Jeff Mitchell and Mirella Lapata, \"Vector-based models of semantic composition\", In Proceedings of the 46th Annual Meeting of ACL: HLT, pages 236-244, Columbus, OH., 2008.", |
| "links": null |
| }, |
| "BIBREF13": { |
| "ref_id": "b13", |
| "title": "Binary codes capable of correcting deletions, insertions, and reversals", |
| "authors": [ |
| { |
| "first": "V", |
| "middle": [], |
| "last": "Levenshtein", |
| "suffix": "" |
| } |
| ], |
| "year": 1966, |
| "venue": "Soviet Physice-Doklady", |
| "volume": "10", |
| "issue": "", |
| "pages": "707--710", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Levenshtein, V, \"Binary codes capable of correcting deletions, insertions, and reversals\", Soviet Physice-Doklady, 10, pages 707-710, 1966.", |
| "links": null |
| }, |
| "BIBREF14": { |
| "ref_id": "b14", |
| "title": "Learning textual entailment using SVMs and string similarity measures", |
| "authors": [ |
| { |
| "first": "Prodromos", |
| "middle": [], |
| "last": "Malakasiotis", |
| "suffix": "" |
| }, |
| { |
| "first": "Ion", |
| "middle": [], |
| "last": "Androutsopoulos", |
| "suffix": "" |
| } |
| ], |
| "year": 2007, |
| "venue": "Proceedings of ACL-PASCAL Workshop on Textual Entailment and Paraphrasing", |
| "volume": "", |
| "issue": "", |
| "pages": "42--47", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Prodromos Malakasiotis, Ion Androutsopoulos, \"Learning textual entailment using SVMs and string similarity measures\", In Proceedings of ACL-PASCAL Workshop on Textual Entailment and Paraphrasing, pages 42-47, Prague, Czech Republic, 2007.", |
| "links": null |
| }, |
| "BIBREF15": { |
| "ref_id": "b15", |
| "title": "Using dependency-based features to take the \"parafarce\" out of paraphrase", |
| "authors": [ |
| { |
| "first": "S", |
| "middle": [], |
| "last": "Wan", |
| "suffix": "" |
| }, |
| { |
| "first": "M", |
| "middle": [], |
| "last": "Dras", |
| "suffix": "" |
| }, |
| { |
| "first": "R", |
| "middle": [], |
| "last": "Dale", |
| "suffix": "" |
| }, |
| { |
| "first": "C", |
| "middle": [], |
| "last": "Paris", |
| "suffix": "" |
| } |
| ], |
| "year": 2006, |
| "venue": "Proceedings of the Australasian Language Technology Workshop", |
| "volume": "", |
| "issue": "", |
| "pages": "131--138", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Wan, S., Dras, M., Dale, R., & Paris, C., \"Using dependency-based features to take the \"parafarce\" out of paraphrase\", In Proceedings of the Australasian Language Technology Workshop, pages 131-138, Sydney, Australia, 2006.", |
| "links": null |
| }, |
| "BIBREF16": { |
| "ref_id": "b16", |
| "title": "BLEU: a method for automatic evaluation of machine translation", |
| "authors": [ |
| { |
| "first": "Kishore", |
| "middle": [], |
| "last": "Papineni", |
| "suffix": "" |
| }, |
| { |
| "first": "Salim", |
| "middle": [], |
| "last": "Roukos", |
| "suffix": "" |
| }, |
| { |
| "first": "Todd", |
| "middle": [], |
| "last": "Ward", |
| "suffix": "" |
| }, |
| { |
| "first": "Wei-Jing", |
| "middle": [], |
| "last": "Zhu", |
| "suffix": "" |
| } |
| ], |
| "year": 2002, |
| "venue": "Proceedings of the 40th Annual Meeting on ACL", |
| "volume": "", |
| "issue": "", |
| "pages": "311--318", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Kishore Papineni, Salim Roukos, Todd Ward and Wei-Jing Zhu, \"BLEU: a method for automatic evaluation of machine translation\", In Proceedings of the 40th Annual Meeting on ACL, pages 311-318, Philadelphia, PA, 2002.", |
| "links": null |
| }, |
| "BIBREF17": { |
| "ref_id": "b17", |
| "title": "Re-evaluating machine translation results with paraphrase support", |
| "authors": [ |
| { |
| "first": "Liang", |
| "middle": [], |
| "last": "Zhou", |
| "suffix": "" |
| }, |
| { |
| "first": "Chin-Yew", |
| "middle": [], |
| "last": "Lin", |
| "suffix": "" |
| }, |
| { |
| "first": "Eduard", |
| "middle": [], |
| "last": "Hovy", |
| "suffix": "" |
| } |
| ], |
| "year": 2006, |
| "venue": "Proceedings of the Conference on EMNLP", |
| "volume": "", |
| "issue": "", |
| "pages": "77--84", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Liang Zhou, Chin-Yew Lin and Eduard Hovy, \"Re-evaluating machine translation results with paraphrase support\", In Proceedings of the Conference on EMNLP, pages 77-84 , Sydney, Australia, 2006.", |
| "links": null |
| }, |
| "BIBREF18": { |
| "ref_id": "b18", |
| "title": "Paraphrase recognition using machine learning to combine similarity measures", |
| "authors": [ |
| { |
| "first": "Prodromos", |
| "middle": [], |
| "last": "Malakasiotis", |
| "suffix": "" |
| } |
| ], |
| "year": 2009, |
| "venue": "Proceedings of the 47th Annual Meeting of ACL and the 4th International Joint Conference on Natural Language Processing of the Asian Federation of Natural Language Processing", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Prodromos Malakasiotis, \"Paraphrase recognition using machine learning to combine similarity measures\", In Proceedings of the 47th Annual Meeting of ACL and the 4th International Joint Conference on Natural Language Processing of the Asian Federation of Natural Language Processing, Suntec, Singapore, 2009.", |
| "links": null |
| }, |
| "BIBREF19": { |
| "ref_id": "b19", |
| "title": "Dependency Syntax: Theory and Practice", |
| "authors": [ |
| { |
| "first": "Igor", |
| "middle": [], |
| "last": "Mel", |
| "suffix": "" |
| }, |
| { |
| "first": "'", |
| "middle": [], |
| "last": "Cuk", |
| "suffix": "" |
| } |
| ], |
| "year": 1987, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Igor Mel'cuk, \"Dependency Syntax: Theory and Practice\", State University of New York Press, 1987.", |
| "links": null |
| }, |
| "BIBREF20": { |
| "ref_id": "b20", |
| "title": "Dependency Parsing", |
| "authors": [ |
| { |
| "first": "Sandra", |
| "middle": [], |
| "last": "K\u00fcbler", |
| "suffix": "" |
| }, |
| { |
| "first": "Ryan", |
| "middle": [], |
| "last": "Mcdonald", |
| "suffix": "" |
| }, |
| { |
| "first": "Joakim", |
| "middle": [], |
| "last": "Nivre", |
| "suffix": "" |
| } |
| ], |
| "year": 2009, |
| "venue": "Synthesis Lectures on Human Language Technologies", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Sandra K\u00fcbler, Ryan McDonald, and Joakim Nivre, \"Dependency Parsing\". Synthesis Lectures on Human Language Technologies. Morgan and Claypool Publishers, 2009.", |
| "links": null |
| }, |
| "BIBREF21": { |
| "ref_id": "b21", |
| "title": "The tree-to-tree editing problem", |
| "authors": [ |
| { |
| "first": "S", |
| "middle": [], |
| "last": "Selkow", |
| "suffix": "" |
| } |
| ], |
| "year": 1977, |
| "venue": "The Journal of Information Processing Letters", |
| "volume": "6", |
| "issue": "6", |
| "pages": "184--186", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Selkow, S., \"The tree-to-tree editing problem\", The Journal of Information Processing Letters, Volume 6, No. 6, 184-186, 1977.", |
| "links": null |
| }, |
| "BIBREF22": { |
| "ref_id": "b22", |
| "title": "The tree-to-tree correction problem", |
| "authors": [ |
| { |
| "first": "Kuo-Chung", |
| "middle": [], |
| "last": "Tai", |
| "suffix": "" |
| } |
| ], |
| "year": 1979, |
| "venue": "The Journal of ACM", |
| "volume": "26", |
| "issue": "3", |
| "pages": "422--433", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Kuo-Chung Tai, \"The tree-to-tree correction problem\", The Journal of ACM, Volume 26, No. 3, 422-433, 1979.", |
| "links": null |
| }, |
| "BIBREF23": { |
| "ref_id": "b23", |
| "title": "Simple fast algorithms for the editing distance between trees and related problems", |
| "authors": [ |
| { |
| "first": "Kaizhong", |
| "middle": [], |
| "last": "Zhang", |
| "suffix": "" |
| }, |
| { |
| "first": "Dennis", |
| "middle": [], |
| "last": "Shasha", |
| "suffix": "" |
| } |
| ], |
| "year": 1989, |
| "venue": "SIAM Journal of Computing", |
| "volume": "18", |
| "issue": "6", |
| "pages": "1245--1262", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Kaizhong Zhang and Dennis Shasha, \"Simple fast algorithms for the editing distance between trees and related problems\", SIAM Journal of Computing, Volume 18, No. 6, pages 1245-1262, 1989.", |
| "links": null |
| }, |
| "BIBREF24": { |
| "ref_id": "b24", |
| "title": "Anaphora resolution in EXTRANS", |
| "authors": [ |
| { |
| "first": "D", |
| "middle": [], |
| "last": "Oll\u00b4a", |
| "suffix": "" |
| }, |
| { |
| "first": "R", |
| "middle": [], |
| "last": "Schwitter", |
| "suffix": "" |
| }, |
| { |
| "first": "F", |
| "middle": [], |
| "last": "Rinaldi", |
| "suffix": "" |
| }, |
| { |
| "first": "J", |
| "middle": [], |
| "last": "Dowdall", |
| "suffix": "" |
| }, |
| { |
| "first": "M", |
| "middle": [], |
| "last": "Hess", |
| "suffix": "" |
| } |
| ], |
| "year": 2003, |
| "venue": "Proc. of the Int. Symposium on Reference Resolution and Its Applications to Question Answering and Summarization", |
| "volume": "", |
| "issue": "", |
| "pages": "23--25", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "oll\u00b4a, D., Schwitter, R., Rinaldi, F., Dowdall, J., & Hess, M, Anaphora resolution in EXTRANS. In Proc. of the Int. Symposium on Reference Resolution and Its Applications to Question Answering and Summarization, pp. 23-25, Venice, Italy,2003.", |
| "links": null |
| }, |
| "BIBREF25": { |
| "ref_id": "b25", |
| "title": "Machine Learning", |
| "authors": [ |
| { |
| "first": "T", |
| "middle": [], |
| "last": "Mitchell", |
| "suffix": "" |
| } |
| ], |
| "year": 1997, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Mitchell, T, \"Machine Learning\", Mc-Graw Hill. 1997.", |
| "links": null |
| }, |
| "BIBREF26": { |
| "ref_id": "b26", |
| "title": "Introduction to Machine Learning", |
| "authors": [ |
| { |
| "first": "Ethem", |
| "middle": [], |
| "last": "Alpaydin", |
| "suffix": "" |
| } |
| ], |
| "year": 2004, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Ethem Alpaydin, \"Introduction to Machine Learning\", The MIT Press, 2004.", |
| "links": null |
| }, |
| "BIBREF27": { |
| "ref_id": "b27", |
| "title": "Paraphrasing spoken Chinese using a paraphrase corpus", |
| "authors": [ |
| { |
| "first": "Yujie", |
| "middle": [], |
| "last": "Zhang", |
| "suffix": "" |
| }, |
| { |
| "first": "Kazuhide", |
| "middle": [], |
| "last": "Yamamoto", |
| "suffix": "" |
| } |
| ], |
| "year": 2005, |
| "venue": "The Journal of Natural Language Engineering", |
| "volume": "11", |
| "issue": "4", |
| "pages": "417--434", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Yujie Zhang and Kazuhide Yamamoto, \"Paraphrasing spoken Chinese using a paraphrase corpus\", The Journal of Natural Language Engineering, Volume 11, No. 4, pages 417-434, December, 2005.", |
| "links": null |
| }, |
| "BIBREF28": { |
| "ref_id": "b28", |
| "title": "Efficient kernels for sentence pair classification", |
| "authors": [ |
| { |
| "first": "Fabio", |
| "middle": [ |
| "Massimo" |
| ], |
| "last": "Zanzotto", |
| "suffix": "" |
| }, |
| { |
| "first": "Lorenzo", |
| "middle": [], |
| "last": "Dell'arciprete", |
| "suffix": "" |
| } |
| ], |
| "year": 2009, |
| "venue": "Proceedings of the 2009 Conference on Empirical Methods in Natural Language Processing", |
| "volume": "1", |
| "issue": "", |
| "pages": "91--100", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Fabio Massimo Zanzotto and Lorenzo Dell'Arciprete, \"Efficient kernels for sentence pair classification\", In Proceedings of the 2009 Conference on Empirical Methods in Natural Language Processing, Volume 1, pages 91-100, Singapore, 2009.", |
| "links": null |
| }, |
| "BIBREF29": { |
| "ref_id": "b29", |
| "title": "Toward finely differentiated evaluation metrics for machine translation", |
| "authors": [ |
| { |
| "first": "Eduard", |
| "middle": [], |
| "last": "Hovy", |
| "suffix": "" |
| } |
| ], |
| "year": 1999, |
| "venue": "Proceedings of the Eagles Workshop on Standards and Evaluation", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Eduard Hovy, \"Toward finely differentiated evaluation metrics for machine translation\", In Proceedings of the Eagles Workshop on Standards and Evaluation, Pisa, Italy, 1999.", |
| "links": null |
| }, |
| "BIBREF30": { |
| "ref_id": "b30", |
| "title": "The ARPA MT evaluation methodologies: evolution, lessons, and future approaches", |
| "authors": [ |
| { |
| "first": "J", |
| "middle": [ |
| "S" |
| ], |
| "last": "White", |
| "suffix": "" |
| }, |
| { |
| "first": "T", |
| "middle": [], |
| "last": "O'connell", |
| "suffix": "" |
| } |
| ], |
| "year": 1994, |
| "venue": "Proceedings of the First Conference of the Association for Machine Translation in the Americas", |
| "volume": "", |
| "issue": "", |
| "pages": "193--205", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "J.S. White and T. O'Connell, \"The ARPA MT evaluation methodologies: evolution, lessons, and future approaches\", In Proceedings of the First Conference of the Association for Machine Translation in the Americas, pages 193-205, Columbia, Maryland, 1994.", |
| "links": null |
| }, |
| "BIBREF31": { |
| "ref_id": "b31", |
| "title": "Chinese Characters Conversion System based on Lookup Table and Language Model", |
| "authors": [ |
| { |
| "first": "Min-Hsiang", |
| "middle": [], |
| "last": "Li", |
| "suffix": "" |
| }, |
| { |
| "first": "Shih-Hung", |
| "middle": [], |
| "last": "Wu", |
| "suffix": "" |
| }, |
| { |
| "first": "Ping-Che", |
| "middle": [], |
| "last": "Yang", |
| "suffix": "" |
| }, |
| { |
| "first": "Tsun", |
| "middle": [], |
| "last": "Ku", |
| "suffix": "" |
| } |
| ], |
| "year": 2010, |
| "venue": "Proceedings of the 22nd Conference on Computational Linguistics and Speech Processing", |
| "volume": "", |
| "issue": "", |
| "pages": "113--127", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Min-Hsiang Li, Shih-Hung Wu, Ping-che Yang and Tsun Ku, \"Chinese Characters Conversion System based on Lookup Table and Language Model\", In Proceedings of the 22nd Conference on Computational Linguistics and Speech Processing (ROCLING 2010), pages 113-127, Nantou, Taiwan, September 2010.", |
| "links": null |
| }, |
| "BIBREF32": { |
| "ref_id": "b32", |
| "title": "Making tree kernels practical for natural language learning", |
| "authors": [ |
| { |
| "first": "Alessandro", |
| "middle": [], |
| "last": "Moschitti", |
| "suffix": "" |
| } |
| ], |
| "year": 2006, |
| "venue": "Proceedings of 11th Conference of the European Chapter of the Association for Computational Linguistics (EACL2006)", |
| "volume": "", |
| "issue": "", |
| "pages": "113--120", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Alessandro Moschitti, \"Making tree kernels practical for natural language learning\", In Proceedings of 11th Conference of the European Chapter of the Association for Computational Linguistics (EACL2006), pages 113-120, Treato, Italy, 2006.", |
| "links": null |
| }, |
| "BIBREF33": { |
| "ref_id": "b33", |
| "title": "ICTCLAS", |
| "authors": [], |
| "year": null, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "ICTCLAS, http://ictclas.org/", |
| "links": null |
| }, |
| "BIBREF35": { |
| "ref_id": "b35", |
| "title": "Is it harder to parse Chinese, or the Chinese Treebank?", |
| "authors": [ |
| { |
| "first": "Roger", |
| "middle": [], |
| "last": "Levy", |
| "suffix": "" |
| }, |
| { |
| "first": "Christopher", |
| "middle": [], |
| "last": "Manning", |
| "suffix": "" |
| } |
| ], |
| "year": 2003, |
| "venue": "Proceeding of the 41st Annual Meeting on Association for Computational Linguistics", |
| "volume": "1", |
| "issue": "", |
| "pages": "439--446", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Roger Levy and Christopher Manning, \"Is it harder to parse Chinese, or the Chinese Treebank?\", In Proceeding of the 41st Annual Meeting on Association for Computational Linguistics, Volume 1, pages 439-446, Sapporo Convention Center, Japan, 2003.", |
| "links": null |
| }, |
| "BIBREF36": { |
| "ref_id": "b36", |
| "title": "LIBSVM", |
| "authors": [], |
| "year": null, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "LIBSVM, http://www.csie.ntu.edu.tw/~cjlin/libsvm/", |
| "links": null |
| } |
| }, |
| "ref_entries": { |
| "TABREF0": { |
| "num": null, |
| "text": "\u6458\u8981 \u6587\u5b57\u860a\u6db5(Textual Entailment)\u7684\u5b9a\u7fa9\u662f\u5224\u65b7\u5169\u500b\u53e5\u5b50\u80fd\u5426\u4e92\u76f8\u63a8\u8ad6\u3002\u63a8\u8ad6\u53ef \u5206\u70ba\u4e94\u7a2e\u985e\u578b\uff1a\u6b63\u5411\u3001\u53cd\u5411\u3001\u96d9\u5411\u3001\u77db\u76fe\u3001\u7368\u7acb\u3002\u9019\u4e94\u7a2e\u985e\u578b\u5206\u5225\u4ee3\u8868\u8457\u4e0d\u540c\u7684 \u860a\u6db5\u95dc\u4fc2\u3002\u6587\u5b57\u860a\u6db5\u8fa8\u8b58(Textual Entailment Recognition)\u662f\u76f8\u7576\u56f0\u96e3\u7684\u81ea\u7136\u8a9e\u8a00 \u3002\u900f\u904e t1 \u7684\u53e5\u5b50\u53ef\u4ee5\u63a8\u8ad6\u51fa\u6db5\u7fa9 s\uff0c\u63a5\u8457\u900f\u904e\u6db5\u7fa9 s \u53ef\u4ee5\u63a8 \u8ad6\u51fa t2 \u4f8b\u5982\uff1at1\uff1a\u5c0f\u660e\u6bba\u4e86\u5c0f\u83ef\u3002t2\uff1a\u5c0f\u83ef\u6b7b\u4e86\u3002\u5f9e t1 \u6211\u5011\u53ef\u4ee5\u63a8\u8ad6\u51fa\u7684 s \u6709\u5f88", |
| "content": "<table><tr><td>\u7684\u8cc7\u8a0a\u662f\u4e92\u76f8\u77db\u76fe\u7684\uff1b\u7368\u7acb\u5247\u662f\u5169\u500b\u53e5\u5b50\u4e2d\u63d0\u5230\u7684\u8cc7\u8a0a\u662f\u5b8c\u5168\u4e0d\u76f8\u95dc\u7684\u3002\u5982\u8868 \u90fd\u4ee5\u7a7a\u767d\u5206\u958b\uff0c\u4e14\u82f1\u6587\u7684\u6587\u6cd5\u4e5f\u5236\u5b9a\u7684\u8f03\u70ba\u6e05\u695a\uff0c\u5982\uff1a\u6642\u614b\u3001\u8a5e\u6027\u2026\u7b49\u7b49\uff0c\u5728\u82f1 1\u3001\u62ec\u865f\u9078\u64c7\u6027\u66ff\u4ee3\uff1a \u4f5c\u8005\u5728\u66f8\u5beb\u6587\u4ef6\u6642\uff0c\u6709\u6642\u6703\u70ba\u4e86\u65b9\u4fbf\uff0c\u800c\u5c07\u4e00\u4e9b\u8a5e\u7c21\u5beb\uff0c\u5730\u540d\u5c31\u662f\u5176\u4e2d\u4e00\u500b \u540c\u4e00\u4ef6\u4e8b\u60c5\uff0c\u53ef\u662f\u65e5\u671f\u4e0d\u4e00\u6a23\u7684\u8a71\uff0c\u9019\u6a23\u5c31\u4e0d\u7b97\u662f\u5b8c\u5168\u5339\u914d\u4e86\u3002 \u4eba\u7684\u773c\u4e2d\u9019\u4e9b\u90fd\u662f\u8868\u9054\u70ba\u7b2c\u4e00\u500b\u5152\u5b50\u7684\u610f\u601d\u3002\u4f46\u662f\u8981\u8b93\u7a0b\u5f0f\u64c1\u6709\u9019\u4e9b\u80cc\u666f\u77e5\u8b58\uff0c \u53cd\u5411\u860a\u6db5(R, Reverse Entailment)\u4e00\u5171\u6709 97 \u500b\u6587\u53e5\u5c0d\uff0c\u96d9\u5411\u860a\u6db5(B, Bidirectional \u6240\u7814\u767c\u7684ICTCLAS\u3002\u5176\u4e2dCKIP\u662f\u7528\u4f86\u8655\u7406\u7e41\u9ad4\u4e2d\u6587\u3002ICTCLAS\u5247\u662f\u7e41\u9ad4\u8207\u7c21\u9ad4 \u8868\u4e03 \u5be6\u9a57\u6240\u4f7f\u7528\u4e4b\u7279\u5fb5 \u8868\u4e5d \u5206\u70ba\u4e94\u985e\u7684 time mapping \u7cfb\u7d71\u7d50\u679c</td></tr><tr><td>\u4e00\u3002 \u8868\u4e00\u4e2d\u96d9\u5411\u860a\u6db5\u7684\u4f8b\u5b50\u6bd4\u8f03\u5c6c\u65bc\u662f\u6539\u5beb(paraphrase)\uff0c\u66f4\u8907\u96dc\u7684\u6587\u5b57\u860a\u6db5\u63a8\u8ad6 \u6587\u4e2d\u90fd\u6709\u8f03\u70ba\u660e\u78ba\u7684\u898f\u7bc4\u3002\u4e2d\u6587\u537b\u662f\u6574\u500b\u53e5\u5b50\u90fd\u9023\u5728\u4e00\u8d77\uff0c\u6240\u4ee5\u7b2c\u4e00\u6b65\u5fc5\u9808\u8981\u5148 \u65b7\u8a5e\uff0c\u65b7\u8a5e\u5728\u4e2d\u6587\u8655\u7406\u4e0a\u9762\u5c31\u6709\u4e00\u5b9a\u7684\u96e3\u5ea6\u3002\u56e0\u70ba\u8655\u7406\u6642\u65b7\u8a5e\u7684\u7d50\u679c\u597d\u58de\u90fd\u6703\u5f71 \u4e00\u500b\u610f\u601d\u53ef\u80fd\u6703\u6709\u5169\u500b\u8a5e\u53ef\u4ee5\u4ee3\u8868\uff0c\u50cf\u662f\u97f3\u8b6f\u3001\u4e2d\u82f1\u6587\u3001\u4ee3\u8868\u6db5\u7fa9\u3001\u7c21\u5beb\u2026 \u7b49\u7b49\u7684\u3002\u5728\u64b0\u5beb\u6587\u7ae0\u6642\uff0c\u70ba\u4e86\u8b93\u8b80\u8005\u53ef\u4ee5\u660e\u78ba\u7684\u77e5\u9053\u4f5c\u8005\u60f3\u8981\u8868\u9054\u7684\u8cc7\u8a0a\uff0c\u4f5c\u8005 \u6642\u5e38\u88ab\u7c21\u5beb\u7684\u5c0d\u8c61\u3002\u4f8b\u5982\uff1a \u300c\u53f0\u7063\u3001\u5370\u5ea6\u3001\u7f8e\u570b\u300d\u9019\u4e9b\u570b\u5bb6\u7684\u540d\u7a31\uff0c\u6642\u5e38\u90fd\u6703\u88ab \u7c21\u5beb\u70ba\uff1a \u300c\u53f0\u3001\u5370\u3001\u7f8e\u300d \u3002\u4f46\u662f\u7576\u6211\u5011\u5728\u8655\u7406\u6587\u5b57\u860a\u6db5\u5206\u6790\u7684\u6642\u5019\uff0c\u6211\u5011\u5fc5\u9808\u8981\u5148 \u8868\u56db \u6279\u914d\u7a0b\u5ea6\u5206\u4f48\u8207\u985e\u5225\u95dc\u4fc2 \u662f\u4e00\u4ef6\u76f8\u7576\u56f0\u96e3\u7684\u4e8b\u60c5\u3002\u56e0\u70ba\u8655\u7406\u4e2d\u6587\u5fc5\u9808\u4e8b\u5148\u7d93\u904e\u65b7\u8a5e\u7cfb\u7d71\u8655\u7406\uff0c\u4f46\u65b7\u8a5e\u7cfb\u7d71 \u6709\u6642\u6703\u628a\u8a5e\u5f59\u65b7\u7684\u592a\u7d30\u3002\u5982\uff1a\u300c\u9577\u7537\u300d\u6703\u88ab\u65b7\u6210\u300c\u9577\u300d\u8207\u300c\u7537\u300d\u5169\u500b\u5b57\u3002\u7d93\u7531\u5256 Entailment)\u4e00\u5171\u6709 82 \u500b\u6587\u53e5\u5c0d\uff0c\u77db\u76fe(C, Contradiction)\u4e00\u5171\u6709 74 \u500b\u6587\u53e5\u5c0d\uff0c\u7368\u7acb (I, Independence)\u4e00\u5171\u6709 81 \u500b\u6587\u53e5\u5c0d\u5982\u8868\u516d\u3002 \u4e2d\u6587\u90fd\u53ef\u4ee5\u8655\u7406\u3002\u6240\u4ee5\u6211\u5011\u7684\u5be6\u9a57\u9078\u64c7\u4f7f\u7528ICTCLAS\u3002 2. Stanford parser[32]\uff1a\u53e6\u5916\u4e00\u500b\u5f88\u91cd\u8981\u7684\u5de5\u5177\u5c31\u5256\u6790\u5668\uff0c\u7531\u65bc\u8981\u8a08\u7b97\u8a9e\u610f\u9700\u8981\u4f7f 1. unigram recall 2. unigram precision Actual Predicted Total F R B I C</td></tr><tr><td>\u5c31\u50cf\u662f \u97ff\u5230\u5f8c\u7e8c\u8655\u7406\u7684\u7d50\u679c\u3002\u4e14\u4e2d\u6587\u7684\u6587\u6cd5\u76f8\u8f03\u65bc\u82f1\u6587\uff0c\u4e5f\u8f03\u70ba\u6a21\u7cca\uff0c\u6240\u4ee5\u4e2d\u6587\u7684\u8a9e\u610f s t \u2192 1 , 2 t \u5224\u65b7\u96e3\u5ea6\u76f8\u8f03\u65bc\u82f1\u6587\u6703\u9ad8\u51fa\u8a31\u591a\u3002 \u6703\u4f7f\u7528\u62ec\u865f\uff0c\u5c07\u8b80\u8005\u53ef\u80fd\u806f\u60f3\u5230\u7684\u8a5e\u4e5f\u90fd\u5305\u542b\u9032\u4f86\uff0c\u907f\u514d\u9020\u6210\u8b80\u8005\u7684\u8aa4\u6703\u3002\u4f8b\u5982\uff1a \u8eca\u8afe\u6bd4\u6838\u4e8b\u6545(\u5207\u723e\u8afe\u8c9d\u5229\u6838\u4e8b\u6545)\u3001\u6e6f\u59c6\u2022\u514b\u9b6f\u65af(Tom Cruise)\u3002\u6240\u4ee5\u5728\u6b64\u6211\u5011\u5c07 \u5c07\u7c21\u5beb\u6062\u5fa9\u6210\u539f\u5730\u540d\u3002\u9019\u6a23\u5728\u4e4b\u5f8c\u6bd4\u5c0d\u624d\u6703\u5339\u914d\u3002 F R I B C \u5b8c\u5168\u5339\u914d 9 17 3 30 17 \u6790\u8655\u7406\u5f8c\uff0c\u7522\u751f\u7684\u5256\u6790\u6a39\u6703\u9577\u5f97\u4e0d\u4e00\u6a23\u3002\u5c0e\u81f4\u8a08\u7b97\u8a9e\u6cd5\u76f8\u4f3c\u5ea6\u6642\u7522\u751f\u51fa\u8a31\u591a\u96dc\u8a0a\u3002 \u70ba\u89e3\u6c7a\u9019\u500b\u554f\u984c\uff0c\u900f\u904e\u4e8b\u5148\u5efa\u7acb\u4e00\u4efd\u540c\u7fa9\u8a5e\u6e05\u55ae\uff0c\u9032\u884c\u540c\u7fa9\u8a5e\u7684\u66ff\u63db\u3002\u4f46\u70ba\u4e86\u6e1b \u8868\u516d \u6bcf\u4e00\u500b\u985e\u5225\u7684\u8cc7\u6599\u6578\u91cf \u6a19\u7c64(Label) \u6578\u91cf(Number) \u7528\u5230 parser\uff0c\u6240\u4ee5\u5728\u6211\u5011\u7684\u5be6\u9a57\u4e2d\u4f7f\u7528\u7684\u662f Stanford parser\u3002\u56e0\u70ba Stanford parser \u662f\u4f9d\u64da Chinese Treebank[37]\u7684\u6a19\u6e96\uff0c\u6240\u4ee5 Stanford parser \u80fd\u5920\u8655\u7406\u82f1\u6587\u4ee5\u53ca\u7c21\u9ad4 3. Bleu precision 4. Bleu recall F 61 3 8 10 5 87 R 0 69 9 14 5 97 s \u2192 \u591a\uff0c\u5982\uff1a\u5c0f\u660e\u662f\u6bba\u4e86\u5c0f\u83ef\u7684\u5147\u624b\u3001\u5c0f\u83ef\u88ab\u6bba\u4e86\u3001\u5c0f\u83ef\u6b7b\u4e86\u3002\u9019\u7a2e\u63a8\u8ad6\u9700\u8981\u6709\u908f\u8f2f \u62ec\u865f\u4e2d\u7684\u6587\u5b57\u8207\u524d\u9762\u7684\u8a5e\u4f7f\u7528\u9663\u5217\u5132\u5b58\u3002\u5728\u7279\u5fb5\u64f7\u53d6\u6642\u6211\u5011\u5c07\u5169\u500b\u8a5e\u90fd\u540c\u6642\u5217\u5165 (\u4e09)\u3001\u8868\u9762\u5b57\u4e32\u7279\u5fb5\u5206\u6790 \u90e8\u5206\u5339\u914d 12 8 7 0 0 \u5c11\u56e0\u65b7\u8a5e\u932f\u8aa4\u9020\u6210\u7684\u5f71\u97ff\uff0c\u6240\u4ee5\u5728\u6b64\u63a1\u7528\u9577\u8a5e\u512a\u5148\u4f86\u4f5c\u66ff\u63db\u3002 \u5411\u524d\u860a\u6db5(F, Forward Entailment) 87 \u4e2d\u6587\u3002 5. Bleu F-measure B 6 6 57 1 12 82</td></tr><tr><td>\u8655\u7406\u554f\u984c\u3002\u7531\u65bc\u4e2d\u6587\u6587\u5b57\u860a\u6db5\u7684\u6587\u737b\u8f03\u7f3a\u4e4f\uff0c\u672c\u7bc7\u8ad6\u6587\u5c07\u4e2d\u6587\u6587\u5b57\u860a\u6db5\u8fa8\u8b58\u63d0\u51fa \u4e86\u4e00\u500b\u6d41\u7a0b\uff0c\u63d0\u4f9b\u7d66\u4e4b\u5f8c\u60f3\u8981\u505a\u9019\u500b\u984c\u76ee\u7684\u4eba\u7684\u4f5c\u70ba\u4e00\u500b\u53c3\u8003\u3002\u4e2d\u6587\u7684\u6587\u5b57\u8655\u7406 \u76f8\u8f03\u65bc\u82f1\u6587\u7684\u6587\u5b57\u8655\u7406\u6709\u8a31\u591a\u4e0d\u540c\u7684\u96e3\u8655\uff0c\u5728\u672c\u7bc7\u8ad6\u6587\u4e2d\uff0c\u6211\u5011\u5c07\u4ecb\u7d39\u8655\u7406\u4e2d\u6587 \u7684\u6587\u5b57\u8655\u7406\u9047\u5230\u7684\u96e3\u8655\u4ee5\u53ca\u8655\u7406\u7684\u6d41\u7a0b\u3002\u6211\u5011\u7684\u7cfb\u7d71\u4f7f\u7528\u652f\u63f4\u5411\u91cf\u6a5f(Support vector machine, SVM)\u4f5c\u70ba\u5340\u5206\u985e\u578b\u7684\u6f14\u7b97\u6cd5\u3002\u4f7f\u7528\u7684\u7279\u5fb5\u5206\u70ba\u5169\u500b\u65b9\u5411\uff1a1.\u6587\u5b57 \u7279\u5fb5 2.\u8a9e\u610f\u7279\u5fb5\u3002 \u95dc\u9375\u5b57\uff1a\u6587\u5b57\u860a\u6db5\u3001tree kernel\u3001\u652f\u6301\u5411\u91cf\u6a5f\u3001\u8a9e\u610f\u5206\u6790 \u4e00\u3001\u7dd2\u8ad6 \u8fd1\u5e7e\u5e74\u4f86\uff0c\u6587\u5b57\u860a\u6db5\u53d7\u5230\u95dc\u6ce8\uff0c\u4e3b\u8981\u662f\u56e0\u70ba\u5927\u5bb6\u77ad\u89e3\u5230\u6587\u5b57\u860a\u6db5\u5c07\u4f7f\u6211\u5011\u80fd \u5920\u66f4\u6e96\u78ba\u7684\u53bb\u63a8\u8ad6\u81ea\u7136\u8a9e\u8a00\u7684\u8a9e\u7fa9\u95dc\u4fc2[1]\u4ee5\u53ca\u8655\u7406\u4e00\u4e9b\u91cd\u8981\u7684\u61c9\u7528[2]\u3002\u50cf\u662f\u6aa2 \u7d22\u7cfb\u7d71\u7d93\u5e38\u6703\u6aa2\u7d22\u51fa\u6210\u5343\u4e0a\u842c\u7b46\u8cc7\u6599\uff0c\u537b\u96e3\u4ee5\u5224\u65b7\u54ea\u500b\u53e5\u5b50\u662f\u8207\u554f\u53e5\u6700\u76f8\u95dc\u7684\u3002 \u65bc\u662f\u53ef\u4ee5\u900f\u904e\u860a\u6db5\u7684\u63a8\u8ad6\uff0c\u5f9e\u9019\u4e9b\u6210\u5343\u4e0a\u842c\u7684\u8cc7\u6599\u4e2d\u6311\u9078\u51fa\u6700\u76f8\u95dc\u7684\u53e5\u5b50\u3002\u7531\u65bc \u5169\u500b\u53e5\u5b50\u4e2d\u7684\u95dc\u4fc2\u6709\u8a31\u591a\u7a2e\uff0c\u4f8b\u5982\uff1a\u860a\u6db5(entailment)\u3001\u6539\u5beb(paraphrase)\u4ee5\u53ca\u7368\u7acb (independence)\u7b49\uff0c\u8a9e\u610f\u63a8\u8ad6\u7684\u76ee\u7684\u5c31\u662f\u5728\u65bc\u5224\u65b7\u5169\u500b\u53e5\u5b50\u4e4b\u9593\u662f\u5c6c\u65bc\u54ea\u4e00\u7a2e\u95dc \u4fc2\u3002\u53ef\u4ee5\u5c07\u63a8\u8ad6\u5206\u70ba\u4e94\u7a2e\u985e\u578b\uff1a\u6b63\u5411\u3001\u53cd\u5411\u3001\u96d9\u5411\u3001\u77db\u76fe\u3001\u7368\u7acb\u9019\u4e94\u7a2e\u985e\u578b\u3002\u9019 \u4e94\u7a2e\u985e\u578b\u4e5f\u5206\u5225\u4ee3\u8868\u8457\u4e0d\u540c\u7684\u860a\u6db5\u95dc\u4fc2\u3002\u6b63\u5411\u63a8\u8ad6\u70ba\u53ef\u4ee5\u5f9e t1 \u53e5\u5b50\u4e2d\u63a8\u8ad6\u51fa t2 \u7684\u53e5\u5b50\uff0c\u5373\u4ee3\u8868 t1 \u53e5\u5b50\u5b8c\u6574\u7684\u5305\u542b\u8457 t2 \u53e5\u5b50\u7684\u8cc7\u8a0a\uff1b\u800c\u53cd\u5411\u63a8\u8ad6\u6b63\u597d\u76f8\u53cd\uff1b\u96d9 \u5411\u5373\u662f t1 \u8207 t2 \u5169\u500b\u53e5\u5b50\u4e92\u76f8\u5b8c\u5168\u5305\u542b\u8457\u5f7c\u6b64\u7684\u8cc7\u8a0a\uff1b\u77db\u76fe\u5373\u662f\u5169\u500b\u53e5\u5b50\u4e2d\u63d0\u5230 \u63a8\u8ad6\u4ee5\u53ca\u8a31\u591a\u80cc\u666f\u77e5\u8b58\u624d\u53ef\u4ee5\u9054\u6210\u3002\u57fa\u65bc\u4e2d\u6587\u8655\u7406\u7684\u6210\u672c\u4ee5\u53ca\u56f0\u96e3\u5ea6\u8003\u91cf\uff0c\u672c\u7bc7 \u8ad6\u6587\u4e3b\u8981\u91dd\u5c0d\u6539\u5beb(paraphrase)\u53bb\u4f5c\u5206\u6790\u3002 \u8868\u4e00 \u5404\u7a2e\u985e\u578b\u7684\u4f8b\u53e5 \u985e\u578b \u4f8b\u53e5 \u6b63\u5411\u860a\u6db5 (forward) t1\uff1a\u65e5\u672c\u6642\u9593 2011 \u5e74 3 \u65e5 11 \u65e5\uff0c\u65e5\u672c\u5bae\u57ce\u7e23\u767c\u751f\u82ae\u6c0f\u898f\u6a21 9.0 \u5f37\u9707\uff0c\u9020\u6b7b\u50b7\u5931\u8e64\u7d04 3 \u842c\u591a\u4eba t2\uff1a\u65e5\u672c\u6642\u9593 2011 \u5e74 3 \u65e5 11 \u65e5\uff0c\u65e5\u672c\u5bae\u57ce\u7e23\u767c\u751f\u82ae\u6c0f\u898f\u6a21 9.0 \u5f37\u9707 \u53cd\u5411\u860a\u6db5 (reverse) t1\uff1a\u7f8e\u570b\u4e3b\u6b0a\u50b5\u4fe1\u8a55\u7d1a\u5f9e\u6700\u9ad8\u7684\uff21\uff21\uff21\u8abf\u964d\u4e00\u7d1a\u5230\uff21\uff21\uff0b t2\uff1a\u7f8e\u570b\u4e3b\u6b0a\u50b5\u4fe1\u8a55\u7d1a\u5f9e\u6700\u9ad8\u7684\uff21\uff21\uff21\u8abf\u964d\u4e00\u7d1a\u5230\uff21\uff21\uff0b\uff0c\u5c07 \u9020\u6210\u7f8e\u570b\u6bcf\u5e74\u7684\u501f\u8cb8\u6210\u672c\u589e\u52a0\u7d04\u4e00\u5343\u5104\u7f8e\u5143 \u96d9\u5411\u860a\u6db5 (bidirection) t1\uff1a\u8cd3\u62c9\u767b\u5728\u5df4\u57fa\u65af\u5766\u7f8e\u8ecd\u653b\u64ca\u4e2d\u6b7b\u4ea1 t2\uff1a\u5df4\u57fa\u65af\u5766\u7f8e\u8ecd\u653b\u64ca\u4e2d\u6bba\u6b7b\u8cd3\u62c9\u767b \u77db\u76fe\u860a\u6db5 (contradiction) t1\uff1a\u5f35\u5b78\u53cb\u5728 1961 \u5e74 7 \u6708 10 \u65e5\uff0c\u751f\u65bc\u9999\u6e2f\uff0c\u7956\u7c4d\u5929\u6d25 t2\uff1a\u5f35\u5b78\u53cb\u751f\u65bc 1960 \u5e74 \u7368\u7acb\u860a\u6db5 (independence) t1\uff1a\u9ece\u59ff\u8207\"\u6b8b\u969c\u5bcc\u8c6a\"\u99ac\u5ef7\u5f37\u7d50\u5a5a\u3002 t2\uff1a\u99ac\u5ef7\u5f37\u70ba\u9999\u6e2f\"\u6771\u65b9\u5831\u696d\u96c6\u5718\"\u5275\u8fa6\u4eba\u4e4b\u4e00\u99ac\u60dc\u5982\u4e4b\u5b50 \u5728\u81ea\u7136\u8a9e\u8a00\u554f\u7b54\u7cfb\u7d71(Question Answering system)\u4e2d\u3002\u4f7f\u7528\u8005\u8f38\u5165\u7684\u662f\u4e00\u500b\u5b8c \u6574\u7684\u554f\u53e5\u3002\u7cfb\u7d71\u56de\u50b3\u7684\u4e5f\u662f\u4e00\u500b\u500b\u5b8c\u6574\u7684\u53e5\u5b50\u3002\u4e26\u4e14\u9023\u7d50\u5230\u76f8\u5c0d\u61c9\u7684\u6587\u7ae0\u4e2d\u3002\u57fa \u672c\u7684\u6aa2\u7d22\uff0c\u901a\u5e38\u6703\u6709\u4e0a\u842c\u53e5\u7684\u53e5\u5b50\u88ab\u7cfb\u7d71\u6311\u9078\u51fa\u4f86\u3002\u8981\u5f9e\u9019\u4e0a\u842c\u500b\u53e5\u5b50\u4e2d\uff0c\u53bb\u6311 \u9078\u51fa\u54ea\u4e00\u500b\u6700\u70ba\u76f8\u95dc\uff0c\u6211\u5011\u5c31\u53ef\u4ee5\u900f\u904e\u6587\u5b57\u860a\u6db5\u7cfb\u7d71\u4f86\u8f14\u52a9\u3002\u4f8b\u5982 ntcir9 \u63d0\u4f9b\u7684 \u7bc4\u4f8b\u4e2d[3]\uff0c\u6211\u5011\u8f38\u5165 t1\uff1a \u300c 1997 \u5e74\u9999\u6e2f\u56de\u6b78\u4e2d\u570b\u3002\u300d \uff0c\u6211\u5011\u53ef\u4ee5\u5f9e\u6aa2\u7d22\u56de\u4f86\u7684\u53e5 \u5b50\u4e2d\uff0c\u6311\u9078\u51fa t2\uff1a \u300c\u9999\u6e2f\u7684\u4e3b\u6b0a\u548c\u9818\u571f\u662f\u5728 1997 \u7531\u82f1\u570b\u653f\u6b0a\u6b78\u9084\u7d66\u4e2d\u570b\u7684\u3002\u300d\u4e26 \u4e14\u5c07\u5b83\u6392\u5e8f\u5230\u8f03\u70ba\u524d\u9762\u7684\u9806\u5e8f\u3002\u56e0\u70ba t1 \u8207 t2 \u7684\u95dc\u4fc2\u70ba\u53cd\u5411\uff0c\u4ee3\u8868\u8457 t1 \u70ba t2 \u860a \u6db5\u610f\u7fa9\u7684\u4e00\u90e8\u5206\u3002\u6240\u4ee5\u6211\u5011\u53ef\u4ee5\u8a8d\u70ba t2 \u5f88\u53ef\u80fd\u662f\u4f7f\u7528\u8005\u8981\u6aa2\u7d22\u7684\u53e5\u5b50\u3002\u6211\u5011\u4e5f \u53ef\u4ee5\u5229\u7528\u5176\u4ed6\u985e\u578b\uff0c\u4f86\u6c7a\u5b9a\u9019\u4e9b\u53e5\u5b50\u6392\u5e8f\u7684\u4f4d\u7f6e\u3002\u50cf\u662f\u8a9e\u610f\u70ba\u77db\u76fe\u3001\u7368\u7acb\u7684\u53e5\u5b50 \u6211\u5011\u5c31\u5c07\u5b83\u6311\u9664\uff0c\u56e0\u70ba\u5b83\u5e7e\u4e4e\u4e0d\u53ef\u80fd\u70ba\u4f7f\u7528\u8005\u6240\u60f3\u8981\u6aa2\u7d22\u7684\u53e5\u5b50\u3002\u6b63\u5411\u3001\u53cd\u5411\u3001 \u96d9\u5411\u6211\u5011\u5c31\u5c07\u4e4b\u6392\u5e8f\u5230\u8f03\u524d\u9762\u7684\u4f4d\u7f6e\uff0c\u56e0\u70ba\u9019\u4e9b\u53e5\u5b50\u8f03\u6709\u53ef\u80fd\u70ba\u4f7f\u7528\u8005\u8981\u6aa2\u7d22\u7684 \u8cc7\u8a0a\u3002 \u76f8\u8f03\u65bc\u82f1\u6587\u4e2d\u6587\u7684\u6587\u5b57\u8655\u7406\u96e3\u5ea6\u9ad8\u51fa\u4e86\u8a31\u591a\u3002\u56e0\u70ba\u5728\u82f1\u6587\u7684\u53e5\u5b50\u4e2d\uff0c\u6bcf\u500b\u8a5e \u8003\u91cf\u3002 1\u3001\u6642\u9593\uff1a \u4e0d\u5339\u914d 1 1 2 0 7 \u53cd\u5411\u860a\u6db5(R, Reverse Entailment) 97 3. LIBSVM [38]\uff1a\u7531\u65bc\u8981\u5206\u985e\u7684\u985e\u5225\u4e00\u5171\u6709\u4e94\u985e\uff0c\u6240\u4ee5\u63a1\u7528 LIBSVM \u4f5c\u70ba\u5206\u985e\u7684 6. difference in sentence length (character) I 19 32 8 17 5 81 \u4e8c\u3001\u76f8\u95dc\u7814\u7a76 2\u3001\u6642\u9593\u6b63\u898f\u5316\uff1a \u8a31\u591a\u6587\u53e5\u5c0d\u4e2d\u7684\u5169\u500b\u53e5\u5b50\u90fd\u542b\u6709\u6642\u9593\u5143\u7d20\uff0c\u7576\u5169\u908a\u6642\u9593\u4e0d\u7b26\u5408\u6642\uff0c\u53ef\u80fd\u662f\u4e00 2\u3001\u8a9e\u6cd5\u5206\u6790(syntax analysis)\uff1a \u96d9\u5411\u860a\u6db5(B, Bidirectional Entailment) 82 \u5206\u985e\u5668\u3002\u56e0\u70ba\u6b64\u5206\u985e\u5668\u53ef\u4ee5\u4e00\u6b21\u5206\u591a\u500b\u985e\u5225\uff0c\u53ef\u4ee5\u907f\u514d\u6389\u50b3\u7d71\u53ea\u80fd\u5206\u5169\u985e\u5c0d\u61c9\u4e0d 7. absolute difference in sentence length (character) C 11 12 19 10 22 74 \u5176\u4ed6\u8a9e\u8a00\u5230\u76ee\u524d\u70ba\u6b62\uff0c\u6587\u5b57\u860a\u6db5\u5728\u4e2d\u6587\u7684\u9818\u57df\u8f03\u7f3a\u5c11\u76f8\u95dc\u7684\u6587\u737b\uff0c\u6211\u5011\u53ea\u80fd \u53c3\u8003\u5176\u4ed6\u8a9e\u8a00\u6587\u5b57\u860a\u6db5\u8655\u7406\u7684\u65b9\u6cd5\u3002\u5728\u82f1\u6587\u8655\u7406\u6587\u5b57\u860a\u6db5\u7684\u6587\u737b[4]\u5c07\u8655\u7406\u82f1\u6587 \u5716\u4e00 \u5169\u53e5\u975e\u5e38\u76f8\u4f3c\u53e5\u5b50\u7684\u4f9d\u5b58\u95dc\u4fc2\u6a39[1] \u5728\u6587\u672c\u4e2d\u6642\u9593\u7684\u8868\u9054\u65b9\u5f0f\u6709\u5f88\u591a\u7a2e\u683c\u5f0f\u53ca\u5b57\u578b\uff0c\u5982\uff1a\u4e2d\u6587\u3001\u6578\u5b57\u5168\u5f62\u3001\u6578\u5b57 \u534a\u5f62\u3001\u6578\u5b57\u4ee5\u300c-\u300d\u9694\u958b\u3001\u7bc4\u570d\u578b\u614b\u7b49\uff0c\u53c3\u898b\u8868\u4e8c\u3002\u5728\u6b64\u5c07\u4ee5\u4e0a\u5404\u7a2e\u683c\u5f0f\uff0c\u7d71\u4e00 \u500b\u5206\u6790\u6587\u5b57\u860a\u542b\u7684\u4f9d\u64da\u3002\u5f9e 421 \u53e5\u6587\u4ef6\u96c6\u4e2d\u6311\u51fa\u4f86\u7684\u6587\u53e5\u5c0d\u4e00\u5171\u6709 146(34.67%) \u500b\u6587\u53e5\u5c0d\u5305\u542b\u6642\u9593\u3002\u6211\u5011\u5c07\u6642\u9593\u5206\u6790\u7d30\u5206\u70ba\u4e09\u500b\u5339\u914d\u7a0b\u5ea6\uff0c\u5982\uff1a\u6642\u9593\u70ba\u5b8c\u5168\u5339\u914d\u3001 2\u3001\u53e5\u5b50\u9577\u5ea6\uff1a \u91dd\u5c0d\u53e5\u5b50\u7684\u9577\u5ea6\u5206\u6790\uff0c\u5c07 t1 \u7684\u53e5\u5b50\u9577\u5ea6\u8207 t2 \u7684\u53e5\u5b50\u9577\u5ea6\u505a\u6bd4\u8f03\u3002\u5373\u5c07 t1 \u7684 \u5206\u6790\u53e5\u6cd5\u9700\u8981\u900f\u904e\u5256\u6790\u5668(parser)\u5c07\u6574\u500b\u53e5\u5b50\u7684\u53e5\u6cd5\u6a19\u6ce8\uff0c\u624d\u53ef\u4ee5\u8a08\u7b97\u6574\u500b \u77db\u76fe(C, Contradiction) 74 \u5230\u985e\u5225\u7684\u554f\u984c\u3002 8. difference in sentence length (term) Total 93 122 101 52 49 421 \u53e5\u5b50\u7684\u53e5\u6cd5\u3002\u672c\u7bc7\u8ad6\u6587\u4f7f\u7528\u7684\u662f\u53f2\u4e39\u4f5b\u5256\u6790\u5668(Stanford parser) [32]\u4f86\u5256\u6790\u53e5\u5b50\u7684 \u7368\u7acb(I, Independence) 81 4.\u7c21\u7e41\u8f49\u63db\u7cfb\u7d71[33]\uff1a\u7531\u65bc Stanford parser \u53ea\u80fd\u5920\u8655\u7406\u7c21\u9ad4\u4e2d\u6587\uff0c\u65bc\u662f\u9700\u8981\u900f\u904e\u4e00 9. absolute difference in sentence length (term) \u6587\u5b57\u860a\u6db5\u7684\u5404\u500b\u65b9\u6cd5\u505a\u4e86\u5206\u6790\uff0c\u4e26\u4e14\u5c07\u5404\u7a2e\u65b9\u6cd5\u6574\u5206\u6210\u4e0b\u9762\u5e7e\u500b\u985e\u5225\u3002 (\u4e00)\u3001\u9700\u8981\u900f\u904e\u80cc\u666f\u77e5\u8b58\u9054\u6210\u4e4b\u65b9\u6cd5 1\u3001\u6574\u5408\u80cc\u666f\u77e5\u8b58\u8207\u908f\u8f2f\u63a8\u8ad6 \u7531\u65bc\u4eba\u5011\u5728\u5e73\u5e38\u751f\u6d3b\u4e2d\u5df2\u7d93\u5f88\u7fd2\u6163\u4f7f\u7528\u81ea\u7136\u8a9e\u8a00\u8868\u9054\u610f\u898b\uff0c\u6240\u4ee5\u81ea\u7136\u7684\u6709\u8a31 \u591a\u80cc\u666f\u77e5\u8b58\u90fd\u5df2\u7d93\u4e0d\u81ea\u89ba\u5f97\u8b8a\u6210\u5e38\u8b58\uff0c\u5224\u65b7\u5169\u53e5\u5b50\u662f\u5426\u53ef\u4ee5\u63a8\u8ad6\u90fd\u89ba\u5f97\u5f88\u7406\u6240\u7576 \u7136\uff0c\u4f46\u662f\u5c0d\u65bc\u96fb\u8166\u4f86\u8aaa\u4e26\u4e0d\u662f\u5982\u6b64\u3002\u63a8\u8ad6\u662f\u53ef\u4ee5\u5f9e\u908f\u8f2f\u860a\u6db5\u4f86\u6aa2\u67e5\uff0c\u50cf\u662f\u4f7f\u7528\u5b9a \u7406\u8b49\u660e\u6587\u53e5\u5c0d\u4e2d\u7684\u6587\u672c\u860a\u6db5[5][6][7][8]\u3002\u6709\u4e00\u90e8\u4efd\u7684\u5b78\u8005\u4f7f\u7528\u542b\u6709\u8a9e\u610f\u7684\u8fad\u5178\u4f86 \u64f7\u53d6\u51fa\u8a5e\u5f59\u7684\u908f\u8f2f\u610f\u7fa9\uff0c\u4f7f\u7528 WordNet[9]\u6216\u8005\u64f4\u5c55 WordNet[10]\u3002\u4f8b\u5982\u65bc WordNet \u4e2d\"\u6697\u6bba\"\u70ba\"\u6bba\"\u7684\u4e0b\u7fa9\u8a5e(\u66f4\u6709\u5177\u9ad4\u7684\u610f\u7fa9) \uff0c\u50cf\u4e0b\u9762\u6240\u793a\uff0cx \u6697\u6bba y \u53ef\u4ee5\u63a8\u8ad6\u6210 \u4e0a\u7fa9\u8a5e x \u6bba y\u3002\u6240\u4ee5\u50cf\u8b00\u6bba\u3001\u523a\u6bba\u2026\u7b49\u7b49\u7684\u9019\u4e9b\u90fd\u53ef\u4ee5\u63a8\u8ad6\u51fa\u76f8\u540c\u7684\u4e0a\u7fa9\u8a5e\u3002 \u2200x\u2200y \u6697\u6bba\u123ax, y\u123b \u21d2 \u6bba\u123ax, y\u123b 2\u3001\u6574\u5408\u80cc\u666f\u77e5\u8b58\u8207\u5411\u91cf\u7a7a\u9593\u6a21\u578b \u5c07\u6bcf\u500b\u8f38\u5165\u8a9e\u8a00\u8868\u9054\u7684\u5b57\uff0c\u5c0d\u6620\u5230\u4e00\u500b\u5411\u91cf\uff0c\u53ef\u4ee5\u770b\u51fa\u7528\u8a5e\u7684\u5206\u4f48\u5f37\u5ea6\uff0c\u7279 \u5225\u662f\u7576\u53e5\u5b50\u5176\u4ed6\u7684\u5b57\u4e5f\u90fd\u5c0d\u61c9\u5230\u540c\u4e00\u500b\u8a9e\u6599\u5eab\u4e2d\uff0c\u5247\u6703\u660e\u986f\u770b\u51fa\u7528\u5b57\u7684\u5206\u4f48 [11] \u3002\u4f8b\u5982\u8981\u6c42\u5171\u540c\u51fa\u73fe\u7684\u5b57\uff0c\u51fa\u73fe\u5728\u7279\u5225\u7684\u8a9e\u6cd5\u4f9d\u5b58\u95dc\u4fc2\u4e0a[12]\u3002\u5728\u6700\u7c21\u55ae\u7684 \u60c5\u6cc1\u4e0b\uff0c\u70ba\u6bcf\u4e00\u500b\u8868\u9054\u5411\u91cf\u7684\u7e3d\u548c\u6216\u5b57\u8a5e\u5c0d\u61c9\u7684\u5411\u91cf\u7e3d\u548c\uff0c\u4f46\u66f4\u8907\u96dc\u7684\u65b9\u6cd5\u4e5f\u5df2 \u63d0\u51fa[13]\u3002\u53e5\u5b50\u53ef\u4ee5\u901a\u904e\u6e2c\u91cf\u6aa2\u6e2c\u8ddd\u96e2\u5411\u91cf\u7684\u5169\u500b\u8f38\u5165\u8868\u9054\u5f0f\u4f86\u5224\u65b7\u662f\u5426\u70ba\u6539\u5beb \u7684\u53e5\u5b50\uff0c\u4f8b\u5982\uff0c\u901a\u904e\u8a08\u7b97\u5176\u9918\u5f26\u76f8\u4f3c\u6027(cosine similarity)\u3002 (\u4e8c)\u3001\u4e0d\u9700\u8981\u900f\u904e\u80cc\u666f\u77e5\u8b58\u9054\u6210\u4e4b\u65b9\u6cd5 1\u3001\u900f\u904e\u8868\u9762\u6587\u5b57 \u5c07\u6587\u5b57\u5c0d\u7d93\u904e\u4e00\u4e9b\u52a0\u5de5\uff0c\u5982\u8a5e\u6027(POS, Part of speech)\u6a19\u8a18\u6216\u547d\u540d\u5be6\u9ad4\u8b58\u5225 (NER, Named entity recognition) \u6a19\u8a18\u3002\u5c0d\u8f38\u5165\u5169\u500b\u5b57\u4e32\u8a08\u7b97\u5b57\u4e32\u7de8\u8f2f\u8ddd\u96e2(edit distance)[14]\uff0c\u8a08\u7b97\u5176\u5171\u540c\u7684\u5b57\u6578\uff0c\u6216\u7d44\u5408\u5e7e\u7a2e\u5b57\u4e32\u76f8\u4f3c\u5ea6\u63aa\u65bd[15]\uff0c\u5305\u62ec\u4f7f\u7528\u6a5f \u5668\u7ffb\u8b6f\u8a55\u6e2c\u7684\u65b9\u6cd5\uff0c\u5982 BLEU(Bilingual Evaluation Understudy) [17][18]\u90fd\u53ef\u80fd\u6709 \u52a9\u65bc\u6587\u5b57\u860a\u6db5\u3002 \u8f49\u63db\u6210\u9663\u5217\u65b9\u5f0f\uff0c\u4ee5\u65b9\u4fbf\u4e4b\u5f8c\u9032\u884c\u6bd4\u5c0d\u7684\u6b65\u9a5f\u3002\u8f49\u63db\u6210\u9663\u5217\u4ee5\u4fbf\u5f8c\u7e8c\u7a0b\u5f0f\u4f5c\u6bd4 \u90e8\u5206\u6642\u9593\u5339\u914d\u3001\u6642\u9593\u5b8c\u5168\u4e0d\u5339\u914d(\u8868\u4e09) \u3002\u5982\u8868\u4e09\u4e2d\uff0c\u6642\u9593\u5b8c\u5168\u6279\u914d\u7684\u4f8b\u5b50\uff0c\u518d \u53e5\u5b50\u9577\u5ea6\u6e1b\u53bb t2 \u7684\u53e5\u5b50\u9577\u5ea6\uff0c\u4ee5\u7528\u4f86\u7d71\u8a08\u5404\u500b\u985e\u5225\u4e2d\uff0c\u53e5\u5b50\u9577\u5ea6\u8207\u985e\u5225\u7684\u95dc\u4fc2\u3002 \u53e5\u6cd5\u3002\u4f46\u7531\u65bc\u53f2\u4e39\u4f5b\u5256\u6790\u5668\u5728\u5256\u6790\u7e41\u9ad4\u4e2d\u6587\u7684\u6a19\u6ce8\u6642\u5e38\u6703\u5224\u65b7\u932f\u8aa4\uff0c\u6240\u4ee5\u4f7f\u7528\u7684 \u500b\u7c21\u7e41\u8f49\u63db\u7684\u7cfb\u7d71\u3002\u9019\u6b21\u5be6\u9a57\u6240\u4f7f\u7528\u7684\u7cfb\u7d71\u70ba\u81ea\u884c\u958b\u767c\u7684\u7cfb\u7d71[33]\u3002 10. Subset tree mapping \u8868\u5341 \u5206\u70ba\u4e94\u985e\u7684 time mapping \u7cfb\u7d71\u7d50\u679c 2\u3001\u57fa\u65bc\u8a9e\u6cd5\u76f8\u4f3c\u5ea6 \u53e6\u4e00\u7a2e\u5e38\u898b\u7684\u65b9\u6cd5\u662f\u5728\u8a9e\u6cd5\u7b49\u7d1a\u3002\u4f9d\u5b58\u8a9e\u6cd5\u5256\u6790\u5668(parser)[20][21]\u666e\u904d\u7528\u65bc \u6587\u672c\u860a\u6db5\u7814\u7a76\uff0c\u4e00\u500b\u53e5\u5b50\u8f38\u51fa\u7684\u5256\u6790\u7d50\u679c\u662f\u4e00\u500b\u5716(\u901a\u5e38\u662f\u4e00\u500b\u6a39\u72c0\u7d50\u69cb) \u5176\u7bc0 \u9ede\u662f\u53e5\u5b50\u7684\u5b57\u6216\u8a5e\u6027\u6a19\u8a18\uff0c\u5176\u908a\u7de3\u5c0d\u61c9\u8a5e\u8207\u8a5e\u4e4b\u9593\u7684\u53e5\u6cd5\u4f9d\u5b58\u95dc\u4fc2\uff0c\u4f8b\u5982\uff1a\u4ee5\u4f9d \u8cf4\u65bc\u52d5\u8a5e\u6216\u540d\u8a5e\u958b\u982d\u7684\u540d\u8a5e\u7247\u8a9e\uff0c\u6216\u8005\u4ee5\u540d\u8a5e\u958b\u982d\u7684\u5f62\u5bb9\u8a5e\u7247\u8a9e\u3002\u5716\u4e00\u986f\u793a\u4e86\u5169 \u53e5\u8a71\u7684\u4f9d\u5b58\u95dc\u4fc2\u6a39[1]\uff0c\u5206\u6790\u9019\u5169\u68f5\u6a39\u53ef\u4ee5\u5f97\u77e5\u9019\u5169\u53e5\u53e5\u5b50\u7684\u860a\u6db5\u95dc\u4fc2\u3002\u4f8b\u5982\u8a08 \u7b97\u5171\u540c\u5256\u6790\u6a39\u7684\u908a[16] [19]\u6216\u4f7f\u7528\u5176\u4ed6\u6a39\u7684\u76f8\u4f3c\u6027\u8a08\u7b97\u65b9\u6cd5\uff0c\u4f8b\u5982\uff1a\u6a39\u7684\u7de8\u8f2f\u8ddd \u96e2[22][23][24]\u3002\u76f8\u4f3c\u6027\u5f97\u5206\u53ef\u4ee5\u8868\u793a\u8f38\u5165\u7684\u53e5\u5b50\u53ef\u80fd\u662f\u6539\u5beb\u7684\u7a0b\u5ea6\u3002 3\u3001\u900f\u904e\u6a5f\u5668\u5b78\u7fd2 \u8a31\u591a\u7cfb\u7d71\u63a1\u7528\u7d50\u5408\u591a\u500b\u6e2c\u91cf\u76f8\u4f3c\u5ea6\u7684\u65b9\u6cd5\uff0c\u5728\u8a08\u7b97\u5404\u7a2e\u7a0b\u5ea6\u7684\u76f8\u4f3c\u5ea6(\u8868\u9762 \u5b57\u4e32\uff0c\u53e5\u6cd5\u548c\u8a9e\u7fa9\u7684\u8868\u793a)\u5408\u4f75\u4f7f\u7528\u6a5f\u5668\u5b78\u7fd2[26][27]\u3002\u6bcf\u4e00\u5c0d\u8f38\u5165\u7684\u6587\u53e5\u5c0d(P1, P2)\uff0c\u7531\u7279\u5fb5\u503c\u5411\u91cf\u4ee3\u8868(f1, . . . , fm) \uff0c\u6211\u5011\u7528\u6a5f\u5668\u5b78\u7fd2\u4f86\u5224\u65b7\u4ed6\u5011\u662f\u5426\u662f\u4e00\u500b\u7279 \u5b9a\u7684\u6539\u5beb\u6216\u6587\u5b57\u860a\u6db5\u3002\u8a72\u5411\u91cf\u5305\u542b\u591a\u500b\u76f8\u4f3c\u5ea6\u7684\u7279\u5fb5\u3002\u524d\u8655\u7406\u968e\u6bb5\u5c07\u6bcf\u500b\u8f38\u5165\u5c0d \u8f49\u63db\u70ba\u4e00\u500b\u7279\u5fb5\u5411\u91cf[28]\u3002\u524d\u8655\u7406\u9084\u5305\u62ec\u6b63\u898f\u5316\uff0c\u4f8b\u5982\uff0c\u65e5\u671f\u5c07\u8f49\u63db\u6210\u4e00\u500b\u7d71\u4e00 \u7684\u683c\u5f0f\uff0c\u500b\u4eba\u7684\u540d\u7a31\uff0c\u7d44\u7e54\uff0c\u5730\u9ede\u7b49\u4f7f\u7528\u547d\u540d\u5be6\u9ad4\u8b58\u5225\u8f49\u63db\u70ba\u6b63\u898f\u5316\u8868\u793a\u3002\u4ee3\u8a5e \u53ca\u6307\u7a31\u8a5e\u8a9e\uff0c\u53ef\u80fd\u6703\u88ab\u66ff\u63db\u7684\u6210\u539f\u672c\u7684\u8a5e[25]\uff0c\u69cb\u53e5\u7684\u5dee\u7570\u4e5f\u53ef\u80fd\u6a19\u6e96\u5316 (\u4f8b\u5982\uff0c \u88ab\u52d5\u53e5\u53ef\u4ee5\u8f49\u63db\u70ba\u4e3b\u52d5\u53e5) \u3002\u7279\u5fb5\u5411\u91cf\u53ef\u4ee5\u7d71\u5305\u4f7f\u7528\u6587\u5b57\u6216\u90e8\u5206\u7684\u53e5\u6cd5\u548c\u8a9e\u7fa9\u8868 \u73fe[29] \u3002\u6700\u5f8c\u9019\u500b\u7279\u5fb5\u5411\u91cf\uff0c\u5c07\u7576\u4f5c\u652f\u6301\u5411\u91cf\u6a5f(SVM)\u7684\u8f38\u5165\u503c\uff0c\u53bb\u5b78\u7fd2\u53ca \u5340\u5206\u5404\u7a2e\u6587\u5b57\u860a\u6db5\u7684\u985e\u5225\u3002 \u4e09\u3001\u8cc7\u6599\u8655\u7406\u8207\u7279\u5fb5\u5206\u6790\uff1a \u672c\u7814\u7a76\u5c07\u8a9e\u610f\u860a\u6db5\u8b58\u5225\u5206\u70ba\u7279\u5fb5\u5206\u6790\u53ca\u6a5f\u5668\u5b78\u7fd2\u5169\u500b\u4e3b\u8981\u90e8\u4efd\u3002\u672c\u7814\u7a76\u63d0\u51fa \u7684\u7279\u5fb5\u5206\u6790\u6d41\u7a0b\u5305\u62ec\u524d\u7f6e\u8655\u7406\u3001\u80cc\u666f\u77e5\u8b58\u7684\u66ff\u63db\u7a0b\u5e8f\u3001\u8868\u9762\u6587\u5b57\u7279\u5fb5\u5206\u6790\u7a0b\u5e8f\u3001 \u8a9e\u610f\u53e5\u6cd5\u5206\u6790\u7a0b\u5e8f\u56db\u500b\u90e8\u5206\u3002\u8cc7\u6599\u4ee5 NTCIR-9 \u63d0\u4f9b\u7684 421 \u5c0d\u6587\u53e5\u5c0d\u70ba\u5206\u6790\u4f9d\u64da [3]\u3002 (\u4e00)\u3001\u524d\u7f6e\u8655\u7406 \u7531\u65bc\u4eba\u5011\u518d\u64b0\u5beb\u53e5\u5b50\u7684\u8868\u9054\u6642\u53ef\u80fd\u6703\u4f7f\u7528\u5230\u4e00\u4e9b\u66ff\u4ee3\u8a5e\u4ee5\u53ca\u7bc4\u570d\u6027\u7684\u8a5e\u6027\u3002 \u7528\u7a0b\u5f0f\u64f7\u53d6\u7279\u5fb5\uff0c\u901a\u5e38\u6703\u9047\u5230\u8655\u7406\u4e0a\u7684\u56f0\u96e3\uff0c\u6240\u4ee5\u6211\u5011\u90fd\u6703\u5fc5\u9808\u8981\u5c07\u8cc7\u6599\u505a\u4e00\u4e9b \u524d\u8655\u7406\uff0c\u624d\u53ef\u4ee5\u7a0b\u5f0f\u9032\u884c\u904b\u7b97\u3002 \u8f03\u3002 3\u3001\u6642\u9593\u904b\u7b97\uff1a \u5728\u6709\u4e9b\u4f8b\u5b50\u4e2d\uff0c\u9700\u8981\u900f\u904e\u904b\u7b97\u4e4b\u5f8c\u624d\u80fd\u5920\u77e5\u9053\u8cc7\u8a0a\u662f\u5426\u5339\u914d\u3002\u4f8b\u5982\uff1at1\uff1a \u300c\u8607 \u54c8\u6258\u653f\u6b0a\u5728\u4e00\u4e5d\u4e5d\u516b\u5e74\u7d50\u675f\uff0c\u57f7\u653f\u5345\u4e8c\u5e74\u3002\u300d \u3001t2\uff1a \u300c\u8607\u54c8\u6258\u4e00\u4e5d\u516d\u516d\u5e74\u57f7\u653f\uff0c\u5c0d \u5370\u5c3c\u9032\u884c\u4e86\u5345\u4e8c\u5e74\u7684\u9435\u8155\u7d71\u6cbb\u3002\u300d\u5728\u4f8b\u5b50\u4e2d t1 \u51fa\u73fe\u6642\u9593\u8a5e\u300c\u4e00\u4e5d\u4e5d\u516b\u5e74\u300d\u8207\u300c\u5345 \u4e8c\u5e74\u300d\u3002\u7d93\u904e\u904b\u7b97\u4e4b\u5f8c\uff0c\u6703\u8207 t2 \u7684\u300c\u4e00\u4e5d\u516d\u516d\u5e74\u300d\u7b26\u5408\uff0c\u6240\u4ee5\u7d93\u904e\u6642\u9593\u904b\u7b97\u7d50 \u679c\uff0ct1 \u7684\u6642\u9593\u5c07\u8207 t2 \u7684\u6642\u9593\u5339\u914d\u3002 \u8868\u4e8c \u5404\u7a2e\u6642\u9593\u8868\u9054\u65b9\u5f0f\u4e4b\u4f8b\u53e5 \u6642\u9593\u578b\u614b \u6642\u9593\u8868\u9054\u65b9\u5f0f \u4e2d\u6587 \u4e00\u4e5d\u4e5d\u4e03\u5e74\u4e8c\u6708\u5eff\u4e09\u65e5 \u6578\u5b57\u5168\u5f62 \uff11\uff19\uff19\uff17\u5e74\uff12\u6708\uff12\uff13\u65e5 \u6578\u5b57\u534a\u578b 1997 \u5e74 2 \u6708 23 \u65e5 \u6578\u5b57\u4ee5\u300c-\u300d\u9694\u958b 1999-05-07 \u7bc4\u570d 1999 \u5e74\u5ef6\u9577\u81f3 2001 \u5e74 (\u4e8c)\u3001\u80cc\u666f\u77e5\u8b58\u7684\u66ff\u63db \u5728\u64b0\u5beb\u6587\u7ae0\u6642\uff0c\u4f5c\u8005\u6703\u4f7f\u7528\u4e00\u4e9b\u7c21\u5beb\u6216\u8005\u662f\u66ff\u4ee3\u8a5e\uff0c\u597d\u8b93\u6574\u7bc7\u6587\u7ae0\u53ef\u4ee5\u66f4\u70ba \u901a\u9806\u3002\u7531\u65bc\u4f5c\u8005\u5df2\u6709\u76f8\u95dc\u7684\u80cc\u666f\u77e5\u8b58\uff0c\u6703\u89ba\u5f97\u9019\u4e9b\u4e8b\u60c5\u662f\u5e38\u8b58\uff0c\u6240\u4ee5\u4e26\u4e0d\u6703\u5c0d\u65bc \u90a3\u4e9b\u66ff\u4ee3\u8a5e\u591a\u505a\u89e3\u91cb\u3002\u4e8b\u5be6\u4e0a\u5f80\u5f80\u8b80\u8005\u95b1\u8b80\u4e00\u4e9b\u6587\u7ae0\u6642\uff0c\u53ef\u80fd\u6703\u56e0\u70ba\u80cc\u666f\u77e5\u8b58\u4e0d \u8db3\u800c\u9700\u8981\u53bb\u67e5\u95b1\u8a31\u591a\u8cc7\u6599\uff0c\u624d\u53ef\u4ee5\u8b80\u61c2\u90a3\u4e9b\u6587\u7ae0\u7684\u610f\u601d\u3002\u9019\u7a2e\u4eba\u5011\u90fd\u6703\u9047\u5230\u7684\u4e8b \u60c5\uff0c\u7a0b\u5f0f\u4e5f\u5fc5\u9808\u8655\u7406\u3002 1\u3001\u5e74\u865f\u7d71\u4e00\uff1a \u5728\u6642\u9593\u7684\u8868\u793a\u8a5e\u4e2d\uff0c\u6709\u4e00\u4e9b\u5e74\u865f\uff0c\u662f\u9700\u8981\u7d93\u904e\u7d71\u4e00\u7684\uff0c\u4f8b\u5982\uff1a\u4e7e\u9686 56 \u5e74\u7b49 2\u3001\u5730\u540d\u6b63\u898f\u5316\uff1a \u6642\u9593\u4e2d\uff0c\u660e\u986f\u7684\u770b\u51fa\uff0c\u6642\u9593\u4e0d\u5339\u914d\u3002 \u8868\u4e09 \u6642\u9593\u6279\u914d\u7a0b\u5ea6\u4e4b\u4f8b\u5b50 \u6279\u914d\u7a0b\u5ea6 \u4f8b\u5b50 \u6642\u9593\u70ba\u5b8c\u5168\u5339\u914d t1\uff1a\u64da\u4ed6\u6240\u77e5\uff0c\u9019\u662f\u67e5\u723e\u65af\u9996\u5ea6\u53c3\u52a0\u96ea\u68a8-\u8377\u82ad\u7279\u5e06\u8239\u8cfd\uff0c\u800c \u67e5\u723e\u65af\u4e00\u5411\u662f\u6ce8\u91cd\u5b89\u5168\u3001\u975e\u5e38\u8b39\u614e\u7684\u4eba\uff0c\u4ed6\u66f4\u60f3\u53c3\u52a0 2000 \u5e74 \u96ea\u68a8\u5967\u904b\u5e06\u8239\u8cfd\u3002 t2\uff1a2000 \u5e74\u5967\u904b\u5728\u96ea\u68a8\u8209\u8fa6 \u90e8\u5206\u6642\u9593\u5339\u914d(1) t1\uff1a\u82e5\u671b\u4fdd\u797f\u4e8c\u4e16\u4e00\u4e5d\u4e03\u516b\u5e74\u5341\u6708\u5341\u516d\u65e5\u88ab\u9078\u70ba\u6559\u5b97 t2\uff1a\u82e5\u671b\u4fdd\u797f\u4e8c\u4e16\u65bc 1978 \u5e74\u7576\u4e0a\u6559\u5b97 \u90e8\u5206\u6642\u9593\u5339\u914d(2) \u7d71\u8a08\u7d50\u679c\u5982\u8868\u4e94\u6240\u793a\u3002 \u8868\u4e94 \u53e5\u5b50\u9577\u5ea6\u8207\u985e\u5225\u95dc\u4fc2 \u985e\u5225 F R I B C \u6578\u91cf 12 -19 -4 0 0 \u7531\u65bc\u5411\u524d\u860a\u6db5(F)\u7684\u5b9a\u7fa9\u70ba\uff0ct1 \u7684\u6db5\u7fa9\u4e2d\u5b8c\u5168\u7684\u5305\u542b\u4e86 t2 \u7684\u53e5\u5b50\uff0c\u6240\u4ee5 t1 \u7684 \u6587\u5b57\u9577\u5ea6\uff0c\u61c9\u8a72\u8981\u6bd4 t2 \u7684\u6587\u5b57\u9577\u5ea6\u9084\u8981\u9577\u8a31\u591a\u3002\u76f8\u53cd\u5730\uff0c\u5982\u679c\u662f\u53cd\u5411\u860a\u6db5(R)\u5247 \u70ba\u53cd\u4e4b\uff0ct2 \u7684\u9577\u5ea6\u61c9\u8a72\u8981\u6bd4 t1 \u7684\u9577\u5f88\u591a\u3002\u800c\u77db\u76fe(C)\u4ee5\u53ca\u96d9\u5411\u860a\u6db5(B)\u7684\u6db5\u7fa9\u5206\u5225 \u70ba\u4e92\u76f8\u77db\u76fe\u4ee5\u53ca\u4e92\u76f8\u5305\u6db5\uff0c\u5728\u53e5\u5b50\u4e2d\u6240\u9700\u63d0\u5230\u7684\u5167\u5bb9\u90fd\u5dee\u4e0d\u591a\u3002\u6240\u4ee5\u53e5\u5b50\u7684\u9577\u5ea6 \u4e5f\u90fd\u6703\u5dee\u4e0d\u591a\u3002\u7368\u7acb(I)\uff0c\u53ea\u8981\u5169\u500b\u53e5\u5b50\u4e0d\u662f\u5728\u8ac7\u540c\u4e00\u500b\u5167\u5bb9\uff0c\u4ed6\u5011\u5c31\u7b97\u662f\u7368\u7acb\u7684 \u53e5\u5b50\u3002\u6240\u4ee5\u53e5\u5b50\u7684\u9577\u5ea6\u70ba\u4e0d\u4e00\u81f4\u3002 3\u3001Bleu(Bilingual Evaluation Understudy)\uff1a Bleu\u7576\u521d\u662f\u88ab\u8a2d\u8a08\u4f86\u6e2c\u91cf\u6a5f\u5668\u7ffb\u8b6f(machine translation)\u7684\u54c1\u8cea\u3002\u4e00\u500b\u826f\u597d\u7684 \u6642\u5019\u628a\u8cc7\u6599\u8f49\u63db\u6210\u7c21\u9ad4\u4e2d\u6587\u6548\u679c\u6703\u8f03\u597d\u3002\u5728\u6b64\u662f\u900f\u904e\u81ea\u884c\u958b\u767c\u51fa\u4f86\u7684\u7c21\u7e41\u8f49\u63db\u7cfb \u7d71\u505a\u8f49\u63db[33]\u3002 (1)\u3001\u8a08\u7b97tree distance \u6709\u8a31\u591a\u5b78\u8005\u4f7f\u7528tree distance\u53bb\u8a08\u7b97\u5169\u500b\u5256\u6790\u6a39\u7684\u76f8\u4f3c\u5ea6[21][22][23]\u3002\u6240\u8b02\u7684 tree distance\u4e3b\u8981\u7684\u6982\u5ff5\u5c31\u662ft1\u7684parser tree\u9700\u8981\u7d93\u904e\u5e7e\u6b21\u63d2\u5165(insert)\u3001\u522a\u9664(delete)\u3001 \u66ff\u4ee3(Substitution)\u624d\u53ef\u4ee5\u7b49\u65bct2\u7684parse tree\u3002\u5982\u5716\u4e8c[24]\u4e2d\uff0c\u53ef\u4ee5\u770b\u51fa\u9019\u5169\u9846\u6a39\u4e2d \u9593\u53ea\u6709\u5dee\u4e86\u4e00\u500bc\u7bc0\u9ede\uff0c\u5728tree distance\u904b\u7b97\u4e2d\u53ea\u9700\u8981\u522a\u9664c\u7bc0\u9ede\u5169\u500bparse tree\u5c31\u5b8c \u5168mapping\u5728\u4e00\u8d77\u3002\u4ed6\u5011\u7684tree distance\u5c31\u662f2\u3002 (2)\u3001 Fast Tree Kernel (FTK) FTK\u70ba\u4fee\u6539Quadratic Tree Kernel (QTK)[5]\u7684\u6f14\u7b97\u6cd5\uff0cQTK\u4e3b\u8981\u7684\u610f\u7fa9\u70ba\u8a08\u7b97 \u5169\u500bTree\u7684\u5339\u914d\u6578\u91cf\u3002FTK\u70ba\u4e8b\u5148\u5c07\u52d5\u8a5e\u70ba\u4e2d\u5fc3\u9ede\uff0c\u5206\u5225\u8ddf\u5176\u4ed6\u7684sub-tree\u5408\u4f75\u6210 subset tree\u3002\u5982\u5716\u4e09\u4e2d\u300c\u5f35\u5b78\u826f\u7684\u7236\u89aa\u662f\u6771\u5317\u8ecd\u95a5\u300d\u5c07\u6703\u4ee5\u662f\u9019\u500b\u52d5\u8a5e\u4f5c\u70ba\u5206\u5272 \u9ede\uff0c\u5206\u5272\u51fa\u300c\u5f35\u5b78\u826f\u7684\u7236\u89aa\u662f\u300d\u3001\u300c\u662f\u6771\u5317\u8ecd\u95a5\u300d\u9019\u5169\u500bsubset tree\u3002\u4e26\u6839\u64da\u6bcf \u5716\u4e09 \u6839\u64da\u52d5\u8a5e\u6240\u5206\u89e3\u51fa\u4f86\u7684sub-set tree \u6587\u53e5\u5c0d \u524d\u8655\u7406 \u65b7\u8a5e \u7279\u5fb5\u64f7\u53d6 (\u4e09)\u3001\u4f7f\u7528\u7279\u5fb5\uff1a 11. Time mapping Actual Predicted Total F R B I C \u4f7f\u7528\u7684 feature \u90e8\u4efd\u662f\u53c3\u8003[16]\u4e2d\u7684\u7279\u5fb5\u3002\u56e0\u70ba[16]\u70ba\u53c3\u8003\u6587\u737b\u4e2d[4]\u6578\u64da\u6700\u597d\u7684\uff0c \u4e94\u3001\u7d50\u8ad6\u8207\u672a\u4f86\u5c55\u671b F 61 3 8 10 5 87 \u4f46\u7531\u65bc[16]\u4e2d\u7684\u7279\u5fb5\u70ba\u8655\u7406\u82f1\u6587\u6642\u4f7f\u7528\uff0c\u6240\u4ee5\u6709\u4e9b\u7279\u5fb5\u4e2d\u6587\u4e26\u6c92\u6709\uff0c\u6240\u4ee5\u53ea\u63a1\u7528 \u90e8\u4efd\u7279\u5fb5\u3002\u4ee5\u53ca\u52a0\u5165\u4e00\u4e9b\u4e0a\u9762\u6240\u5206\u6790\u7684\u90e8\u4efd\u7279\u5fb5\u3002\u6240\u4f7f\u7528\u7684\u7279\u5fb5\u5982\u8868\u4e03\u4e2d\u6240\u5217\u3002 \u76ee\u524d\u5be6\u9a57\u662f\u5c07\u5206\u70ba\u4e94\u7a2e\u985e\u5225\u505a\u56db\u500b\u5be6\u9a57\u3002 1.baseline \u7cfb\u7d71 \u5176\u4e2d baseline \u7cfb\u7d71\u5206\u5225\u4f7f\u7528\u8868\u4e03\u4e2d 1 \u5230 9 \u7684\u7279\u5fb5\u503c\uff0c\u56e0\u70ba\u9019\u4e9b\u7279\u5fb5\u503c\u662f\u8f03\u5bb9\u6613\u505a \u8a08\u7b97\u7684\uff0c\u5be6\u9a57\u7d50\u679c\u5982\u8868\u516b\u6240\u793a\uff0c\u5be6\u9a57\u7684\u6587\u53e5\u5c0d\u4e00\u5171\u6709 421 \u5c0d\u5176\u4e2d\u6709 220 \u5c0d\u662f\u5224\u65b7 R 0 69 9 14 5 97 \u672c\u7bc7\u8ad6\u6587\u63d0\u51fa\u4e86\u4e00\u500b\u8655\u7406\u4e2d\u6587\u6587\u5b57\u860a\u6db5\u8fa8\u8b58\u7684\u6d41\u7a0b\uff0c\u53ef\u4ee5\u7d66\u4e4b\u5f8c\u60f3\u8981\u505a\u4e2d\u6587 B 6 6 57 1 12 82 \u6587\u5b57\u860a\u6db5\u7684\u4eba\u4f5c\u70ba\u4e00\u500b\u53c3\u8003\u3002\u672c\u6b21\u5be6\u9a57\u53ea\u6709\u4f7f\u7528\u90e8\u4efd\u7684\u7279\u5fb5\uff0c\u5be6\u9a57\u7684\u91cf\u4e5f\u4e26\u4e0d\u5927\uff0c I 19 32 8 17 5 81 \u4f46\u662f\u6211\u5011\u5c07\u9019\u6b21\u7684\u7cfb\u7d71\u8996\u70ba\u4e00\u500b baseline\u3002\u53ef\u4ee5\u5f9e\u5be6\u9a57\u4e2d\u770b\u51fa tree mapping \u8207 time C 11 12 19 10 22 74 mapping \u78ba\u5be6\u53ef\u4ee5\u7565\u70ba\u589e\u52a0\u5224\u65b7\u6587\u5b57\u860a\u6db5\u7684\u6548\u679c\u3002\u4e26\u4e14\u5c07\u9577\u5ea6\u7684\u7279\u5fb5\u62ff\u6389\u4e4b\u5f8c\uff0c Total 90 129 100 57 45 421 \u96d6\u7136\u6709\u7565\u70ba\u7684\u4e0b\u964d\uff0c\u4f46\u662f\u5e45\u5ea6\u4e0d\u81f3\u65bc\u5f71\u97ff\u592a\u591a\uff0c\u6240\u4ee5\u4ee5\u6b64\u9a57\u8b49\u4e86\u6b64\u7cfb\u7d71\u4e26\u4e0d\u662f\u975e \u7c21\u7e41\u8f49\u63db \u5256\u6790\u53e5\u5b50 \u6b63\u78ba(52.25%)\u3002 2.tree mapping \u7cfb\u7d71 \u76f8\u8f03\u65bc baseline \u7cfb\u7d71\uff0c\u984d\u5916\u52a0\u5165\u4e86\u7b2c 10 \u500b tree mapping \u7684\u7279\u5fb5\u503c\uff0c\u6b64\u7279\u5fb5\u662f\u8a08\u7b97 Subset Tree mapping \u7684\u503c\u3002\u6240\u4ee5\u4f7f\u7528\u4e86 1\uff5e10 \u7684\u7279\u5fb5\u503c\u3002\u6b64\u5be6\u9a57\u7528\u4f86\u6e2c\u8a66\u770b\u770b\u8a9e \u6cd5\u7279\u5fb5\u5c0d\u65bc\u6587\u5b57\u860a\u6db5\u80fd\u5920\u6709\u591a\u5927\u7684\u5e6b\u52a9\uff0c\u5be6\u9a57\u7d50\u679c\u5982\u8868\u4e5d\u6240\u793a\uff0c\u5be6\u9a57\u7684\u6587\u53e5\u5c0d\u4e00 \u5171\u6709 421 \u5c0d\u5176\u4e2d\u6709 226 \u5c0d\u662f\u5224\u65b7\u6b63\u78ba(53.68%)\u3002 \u5e38\u7684\u4f9d\u8cf4\u9577\u5ea6\u7279\u5fb5\u3002 \u8868\u5341\u4e00 \u5206\u70ba\u4e94\u985e\u7684 time mapping \u7cfb\u7d71\u7d50\u679c \u4e4b\u5f8c\u6211\u5011\u5c07\u589e\u52a0\u5404\u7a2e\u7279\u5fb5\uff0c\u4e26\u4e14\u53bb\u5206\u6790\u51fa\u6709\u7528\u7684\u7279\u5fb5\u4ee5\u6539\u9032\u4e2d\u6587\u6587\u5b57\u860a\u6db5\u7684 \u6548\u679c\u3002\u50cf\u662f\u5e74\u865f\u7d71\u4e00\u3001\u540c\u7fa9\u8a5e\u66ff\u63db\u3001\u5730\u540d\u6b63\u898f\u5316\uff0c\u9019\u4e00\u985e\u9700\u8981\u4e8b\u5148\u5efa\u7acb\u51fa\u80cc\u666f\u77e5 Actual Predicted Total F R B I C \u8b58\u624d\u80fd\u5920\u4f7f\u7528\u7a0b\u5f0f\u53bb\u57f7\u884c\u3002\u4f46\u662f\u9019\u8981\u5efa\u7acb\u51fa\u9019\u4e9b\u80cc\u666f\u77e5\u8b58\u9700\u8981\u82b1\u8cbb\u975e\u5e38\u591a\u7684\u6210\u672c\uff0c F 54 6 12 12 5 87 \u9700\u8981\u7d93\u904e\u5f88\u9577\u7684\u6642\u9593\u53bb\u7d2f\u7a4d\u8a5e\u5f59\u7684\u6578\u91cf\u3002\u76ee\u524d\u6211\u53ea\u6709\u91dd\u5c0d\u8a9e\u6599\u5eab\u4e2d\u9047\u5230\u9700\u8981\u8655\u7406 \u7684\u5efa\u7acb\u51fa\u4f86\uff0c\u6240\u4ee5\u672a\u4f86\u6211\u5011\u5e0c\u671b\u53ef\u4ee5\u5c07\u6bcf\u6b21\u9047\u5230\u8981\u8655\u7406\u7684\u8a5e\uff0c\u5efa\u7acb\u51fa\u8d8a\u4f86\u8d8a\u5b8c\u6574 R 7 58 10 10 5 97 \u6a5f\u5668\u7ffb\u8b6f\u9700\u8981\u5305\u542b\u9069\u7576\uff0c\u6e96\u78ba\u4ee5\u53ca\u6d41\u66a2\u7684\u7ffb\u8b6f[30][31]\u3002Bleu\u662f\u8003\u91cf\u53e5\u5b50\u7684\u76f8\u4f3c \u5ea6\uff0c\u7d93\u904e\u9069\u7576\u4fee\u6539\u53c3\u8003\uff0c\u4e00\u5b9a\u7a0b\u5ea6\u8a5e\u8a9e\u7684\u5dee\u7570\u5728\u9078\u64c7\u548c\u8a9e\u5e8f\u4e0a\u9762\u3002\u800cBleu\u4e3b\u8981\u7684 \u6982\u5ff5\u662f\u4f7f\u7528\u7247\u8a9e\u5339\u914d\u9577\u5ea6\u5e73\u5747\u6b0a\u91cd\u503c\u3002 \u500b\u96c6\u5408\u8a08\u7b97subset tree\u7684\u5339\u914d\u6578\u91cf\uff0c\u4e26\u4e14\u6c42\u51fa\u53c3\u6578\u6700\u5927\u503c S i t \u2208 = max arg [34]\u3002 3.time mapping B 8 2 59 3 10 82 \u7684\u4e00\u500b\u53ef\u4ee5\u8b93\u96fb\u8166\u4f7f\u7528\u7684\u80cc\u666f\u77e5\u8b58\u8fad\u5178\u3002 \u56db\u3001\u5be6\u9a57 \u672c\u6b21\u91dd\u5c0d\u4e94\u7a2e\u985e\u5225\u9032\u884c\u5206\u6790\u8207\u5be6\u9a57\uff0c\u4e26\u4e14\u4f7f\u7528\u4ea4\u53c9\u9a57\u8b49\u7684\u65b9\u5f0f\u4f86\u5be6\u505a\u7684\u7cfb\u7d71\u3002 \u5728\u6211\u5011\u5148\u524d\u7684\u8a66\u9a57\u5206\u6790\u4e2d\uff0c\u53ef\u4ee5\u5f9e\u8868\u4e03\u4e2d\u770b\u51fa\u6642\u9593\u6279\u914d\u662f\u4e00\u500b\u883b\u6709\u7528\u7684\u7279\u5fb5\uff0c\u6240 I 16 37 9 16 5 81 SVM \u4ee5\u55ae\u7368\u5c07\u6b64\u7279\u5fb5\u6311\u9078\u51fa\u4f86\u505a\u5be6\u9a57\u770b\u6548\u679c\u5982\u4f55\u3002\u76f8\u8f03\u65bc baseline \u7cfb\u7d71\u4e2d\uff0c\u800c\u5916\u52a0\u5165 \u8868\u516b \u5206\u70ba\u4e94\u985e\u7684 baseline \u7cfb\u7d71\u7d50\u679c C 18 12 18 8 22 74 t1\uff1a\u8607\u54c8\u6258 1921 \u5e74 6 \u6708 8 \u65e5\u51fa\u751f t2\uff1a\u8607\u54c8\u6258(Suharto\uff0c\u6c11\u9593\u5e38\u7528\u300cSoeharto\u300d \uff0c1921 \u5e74 6 \u6708 8 \u65e5-2008 \u5e74 1 \u6708 27 \u65e5) \u6642\u9593\u5b8c\u5168\u4e0d\u5339\u914d t1\uff1a\u5f35\u85dd\u8b00 1987 \u5e74\u4ee5\u300c\u7d05\u9ad8\u7cb1\u300d\u62ff\u4e0b\u67cf\u6797\u5f71\u5c55\u91d1\u718a\u734e t2\uff1a\u67cf\u6797\u96fb\u5f71\u7bc0\u61c9\u8a72\u662f\u5f35\u85dd\u8b00\u7684\u798f\u5730\u30021988 \u5e74\uff0c\u4ed6\u57f7\u5c0e\u7684\u300a\u7d05 \u9ad8\u7cb1\u300b\u8d0f\u5f97\u4e86\u6700\u4f73\u5f71\u7247\u91d1\u718a\u734e\uff0c\u6210\u70ba\u4e2d\u570b\u96fb\u5f71\u7684\u9996\u500b\u91d1\u718a\u734e \u7d93\u904e\u7d71\u8a08\uff0c\u9019\u4e09\u7a2e\u5339\u914d\u7a0b\u5ea6\u6240\u5305\u542b\u7684\u985e\u5225\u6578\u91cf\u5982\u8868\u56db\uff0c\u53ef\u4ee5\u5f9e\u8868\u4e2d\u770b\u51fa\uff0c\u5b8c \u5168\u5339\u914d\u7684\u90e8\u5206\uff0c\u5c6c\u65bc B \u7684\u6a5f\u7387\u9ad8\u4e86\u5f88\u591a\u3002\u9019\u662f\u56e0\u70ba\u7576\u5169\u500b\u53e5\u5b50\u7684\u610f\u7fa9\u53ef\u4ee5\u4e92\u76f8 \u63a8\u8ad6\u6642\uff0c\u5728\u6642\u9593\u4e0a\u5fc5\u9808\u8981\u5b8c\u5168\u5339\u914d\u624d\u6709\u53ef\u80fd\u610f\u7fa9\u662f\u4e00\u6a23\u7684\u3002\u5982\u679c\u5169\u500b\u53e5\u5b50\u90fd\u5728\u8b1b 4\u3001\u5426\u5b9a\u8a5e\u5075\u6e2c \u7576\u6587\u53e5\u5c0d\u53ea\u6709\u5dee\u5225\u4e00\u500b\u5426\u5b9a\u8a5e\u6642\uff0c\u5f9e\u6587\u5b57\u5c64\u9762\u53bb\u8a08\u7b97\u6587\u53e5\u5c0d\u5f97\u76f8\u4f3c\u5ea6\u6703\u5f97\u5230 \u5f88\u9ad8\u7684\u5206\u6578\u3002\u4f46\u662f\u5f9e\u8a9e\u610f\u5c64\u9762\u4f86\u770b\u6b64\u6587\u53e5\u5c0d\u537b\u5b8c\u5168\u4e0d\u76f8\u540c\u3002\u5982\uff1a \u300c\u4eca\u5929\u5929\u6c23\u5f88\u597d\u300d \u8207\u300c\u4eca\u5929\u5929\u6c23\u5f88\u4e0d\u597d\u300d\u3002\u6b64\u6587\u53e5\u5c0d\u4e2d\u53ea\u6709\u5dee\u5225\u4e00\u500b\u300c\u4e0d\u300d\u5426\u5b9a\u8a5e\u3002\u7136\u800c\u5b8c\u5168\u6539\u8b8a \u4e86\u6b64\u8a9e\u53e5\u7684\u8a9e\u610f\u3002\u6240\u4ee5\u6211\u5011\u63d0\u51fa\u8981\u5075\u6e2c\u6587\u53e5\u5c0d\u4e2d\u662f\u5426\u6709\u5426\u5b9a\u8a5e\uff0c\u4f5c\u70ba\u7279\u5fb5\u3002 (\u56db)\u3001\u8a9e\u610f\u53e5\u6cd5\u5206\u6790\uff1a 1\u3001\u540c\u7fa9\u8a5e\u66ff\u63db\uff1a \u8868\u9054\u540c\u4e00\u500b\u610f\u601d\u7684\u8a5e\u5f59\u6709\u8a31\u591a\uff0c\u4f8b\u5982\uff1a\u5927\u5152\u5b50\u3001\u9577\u7537\u3001\u7372\u5f97\u3001\u5f97\u5230\u2026\u7b49\u3002\u5728 \u5716\u4e8c parse tree\u5c0d\u61c9\u5716 \u6d41\u7a0b\u5982\u5716\u56db\u6240\u793a\uff0c\u9996\u5148\u8f38\u5165\u6587\u53e5\u5c0d\uff0c\u9032\u884c\u524d\u8655\u7406\uff0c\u5176\u4e2d\u524d\u8655\u7406\u5305\u62ec\uff1a\u62ec\u865f\u9078\u64c7\u6027 \u66ff\u4ee3\uff0c\u5e74\u865f\u7d71\u4e00\uff0c\u4ee5\u53ca\u5730\u540d\u6b63\u898f\u5316\u3002\u4e4b\u5f8c\u4f7f\u7528 ICTCLAS \u7cfb\u7d71\u5c0d\u53e5\u5b50\u9032\u884c\u65b7\u8a5e\uff0c \u63a5\u8457\u9032\u884c\u7c21\u7e41\u8f49\u63db\u624d\u5c07\u53e5\u5b50\u4f7f\u7528 Stanford parser \u5c07\u53e5\u5b50\u5256\u6790\uff0c\u63a5\u8457\u9032\u884c\u7279\u5fb5\u64f7\u53d6 \u7684\u52d5\u4f5c\uff0c\u672c\u6b21\u64f7\u53d6\u7684\u7279\u5fb5\u5982\u8868\u4e03\u6240\u793a\uff0c\u4e26\u5c07\u64f7\u53d6\u7684\u7279\u5fb5\u8f38\u5165\u7d66 SVM \u9032\u884c\u8a13\u7df4\u4ee5 \u53ca\u6e2c\u8a66\uff0c\u8f38\u51fa\u5c07\u5f97\u5230\u5224\u65b7\u662f\u5426\u70ba\u860a\u6db5\u7684\u7d50\u679c\u3002 (\u4e00)\u3001\u8cc7\u6599\u4f86\u6e90\uff1a \u672c\u6b21\u7814\u7a76\u6211\u5011\u7684\u8cc7\u6599\u4f86\u6e90\u53d6\u81ea\u65e5\u672c NTCIR \u7b2c\u4e5d\u5c46\u4e2d\uff0cRITE( Recognizing Inference in Text)\u6bd4\u8cfd\u5b50\u9805\u76ee\u7684\u958b\u767c\u8cc7\u6599(Development Data)\u3002\u800c\u5728\u6b64\u8cc7\u6599\u4e2d\uff0c\u4e00 \u904d\u7684\u65b7\u8a5e\u5de5\u5177\u6709\u5169\u7a2e\uff1a\u7531\u4e2d\u592e\u7814\u7a76\u9662\u6240\u7814\u767c\u7684CKIP\u65b7\u8a5e\u7cfb\u7d71[36]\uff0c\u8207\u4e2d\u570b\u79d1\u5b78\u9662 \u5171\u6709 421 \u500b\u6587\u53e5\u5c0d\u3002\u5176\u4e2d\u5411\u524d\u860a\u6db5(F, Forward Entailment)\u4e00\u5171\u6709 87 \u500b\u6587\u53e5\u5c0d\uff0c \u5716\u56db \u672c\u6b21\u5be6\u9a57\u4e4b\u6d41\u7a0b\u5716 (\u4e8c)\u3001\u4f7f\u7528\u5de5\u5177\uff1a 1. ICTCLAS[35]:\u7531\u65bc\u4e2d\u6587\u7684\u53e5\u5b50\u4e26\u4e0d\u50cf\u82f1\u6587\u7684\u53e5\u5b50\uff0c\u6bcf\u500b\u8a5e\u90fd\u4ee5\u7a7a\u767d\u5206\u958b\u3002\u6240 \u4ee5\u8981\u8655\u7406\u4e2d\u6587\u7684\u53e5\u5b50\u9996\u5148\u7b2c\u4e00\u500b\u6b65\u9a5f\u9700\u8981\u9032\u884c\u65b7\u8a5e\uff0c\u5c07\u6bcf\u500b\u8a5e\u5206\u958b\u3002\u76ee\u524d\u8f03\u70ba\u666e \u8f38\u51fa\u5224\u65b7\u985e\u5225 \u7b2c 11 \u500b time mapping \u7684\u7279\u5fb5\uff0c\u6240\u4ee5\u4f7f\u7528\u4e86 1\uff5e9 \u7684\u4e26\u4e14\u984d\u5916\u52a0\u5165\u4e86\u7b2c 11 \u500b\u7279\u5fb5\uff0c \u5be6\u9a57\u7d50\u679c\u5982\u8868\u5341\u6240\u793a\uff0c\u5be6\u9a57\u7684\u6587\u53e5\u5c0d\u4e00\u5171\u6709 421 \u5c0d\u5176\u4e2d\u6709 223 \u5c0d\u662f\u5224\u65b7\u6b63\u78ba (52.96%)\u3002 4. Remove length \u6211\u5011\u53ef\u4ee5\u5f9e\u8868\u4e94\u4e2d\u770b\u51fa\u672c\u8cc7\u6599\u96c6\u7684\u9577\u5ea6\u7279\u5fb5\u8f03\u70ba\u660e\u986f\uff0c\u7576\u7136\u9019\u6709\u53ef\u80fd\u662f\u672c\u8cc7\u6599\u96c6 \u624d\u6709\u7684\u7279\u8272\uff0c\u6240\u4ee5\u5728\u6b64\u5c07\u7b2c 6 \u5230 9 \u500b\u95dc\u65bc\u9577\u5ea6\u7279\u5fb5\u62ff\u6389\u3002\u7528\u4f86\u6e2c\u8a66\u9019\u6b21\u5be6\u9a57\u662f\u5426 Total 93 124 103 52 49 421 \u6709 209 \u5c0d\u662f\u5224\u65b7\u6b63\u78ba(49.64%)\u3002 C 11 12 21 10 20 74 \u904e\u5ea6\u4f9d\u8cf4\u9577\u5ea6\u7279\u5fb5\u3002\uff0c\u5be6\u9a57\u7d50\u679c\u5982\u8868\u5341\u4e00\u6240\u793a\uff0c\u5be6\u9a57\u7684\u6587\u53e5\u5c0d\u4e00\u5171\u6709 421 \u5c0d\u5176\u4e2d Actual Total 99 115 108 59 40 421 Predicted Total F R B I C F 60 3 9 10 5 Acknowledgement 87 R 0 68 9 15 5 \u672c\u7814\u7a76\u4f9d\u7d93\u6fdf\u90e8\u88dc\u52a9\u8ca1\u5718\u6cd5\u4eba\u8cc7\u8a0a\u5de5\u696d\u7b56\u9032\u6703 \u300c100 \u5e74\u5ea6\u6578\u4f4d\u532f\u6d41\u670d\u52d9\u958b\u653e\u5e73\u53f0 97 B 5 6 56 1 14 \u6280\u8853\u7814\u767c\u8a08\u756b\u300d\u8fa6\u7406\u3002\u611f\u8b1d\u570b\u79d1\u6703\u8d0a\u52a9\u90e8\u5206\u7814\u7a76\u7d93\u8cbb\uff0c\u8a08\u756b\u7de8\u865f NSC-82 I 17 35 8 16 5 81 100-2221-E-324 -025-MY2\u3002</td></tr></table>", |
| "html": null, |
| "type_str": "table" |
| } |
| } |
| } |
| } |
